Digital analyte analysis

Abstract

To provide droplet based digital PCR and methods for analyzing a target nucleic acid using the same.SOLUTION: A method includes the steps of: sectioning a sample including a plurality of target nucleic acids into a plurality of sections; amplifying a first target nucleic acid; and detecting a first marker and a second marker at least in the first section of the plurality of sections. Therein at least a first section of the plurality of sections includes: the first target nucleic acid; and a primer / probe mix which includes a first probe configured to bind to the nucleic acid and to have a first marker, a second probe configured to bind to a second target nucleic acid and to have a second marker, a third probe configured to bind to a third target nucleic acid and to have the first marker, a fourth probe configured to bind to the third target nucleic acid and to have the second marker, and first, second and third primer pairs configured to respectively bind to the first, second and third target nucleic acid, and in which the second marker is different from the first marker.SELECTED DRAWING: Figure 1

Description

[Technical field]

[0001] REFERENCE TO RELATED APPLICATIONS This application claims priority to U.S. Provisional Application No. 61 / 388,937, filed October 1, 2010, U.S. Provisional Application No. 61 / 347,158, filed May 21, 2010, U.S. Provisional Application No. 61 / 331,490, filed May 5, 2010, and U.S. Provisional Application No. 61 / 304,163, filed February 12, 2010, the contents of each of which are incorporated herein by reference in their entirety.

[0002] FIELD OF THEINVENTION The present invention relates generally to droplet-based digital PCR and methods for using same to analyze target nucleic acids. [Background technology]

[0003] Assays that rely on analyzing nucleic acid molecules from bodily fluids for the presence of mutations have been developed, thus leading to early diagnosis of certain diseases, such as cancer. However, in a typical bodily fluid sample, any abnormal nucleic acid containing a mutation of interest is often present in small amounts (e.g., less than 1%) relative to the total amount of nucleic acid in the bodily fluid sample. This can result in failure to detect the small amount of abnormal nucleic acid due to stochastic sampling bias.

[0004] The advent of PCR and real-time PCR methodologies has greatly improved the analysis of nucleic acids, both in terms of throughput and in terms of quantity. While traditional PCR techniques generally rely on end-point, and sometimes semi-quantitative, analysis of amplified DNA targets by agarose gel electrophoresis, real-time PCR (or qPCR) methods are adapted for precise quantification of exponential amplification as the reaction progresses. qPCR reactions are monitored using a variety of highly sequence-specific fluorescent probe technologies or by using non-specific DNA-intercalating fluorogenic dyes.

[0005] As the need for higher throughput in analyzing multiple targets in parallel continues to grow in genomics and genetics, as well as the need for more efficient use of samples in medically related fields such as diagnostics, the ability to simultaneously perform and quantify multiple amplifications in the same reaction volume (multiplexing) is paramount for both PCR and qPCR. While end-point PCR can accommodate high levels of amplicon multiplexing, such ample capacity for multiplexing probe-based qPCR reactions remains difficult to achieve for a number of reasons. For example, most commercial real-time thermal cyclers can only accommodate up to four different detection fluorophores as a result of the limited spectral resolution of common fluorophores, translating to a 4× multiplexing capacity. In addition, while optimization of single-target primer / probe reactions is now standard practice, the use of primers and probes in combination for multiple reactions has the potential to alter thermodynamic efficiencies and / or chemical kinetics, necessitating extensive troubleshooting and optimization. Very high degrees of multiplexing, greater than 100×, have been demonstrated with "sloppy" molecular beacons and "one of many" detection formats for pathogen identification using melting points as fingerprints, but this approach is limited to applications where the likelihood of the presence of multiple simultaneous targets is only small. A semi-multiplexing method achieved 19× in a two-step reaction using a common multiplex preamplification in the first step followed by another singleplex quantitative PCR in the second step. However, a universal single-pot solution for qPCR multiplexing does not yet exist.

[0006] Digital PCR (dPCR) is an alternative quantification method that divides a dilute sample into many separate reactions. See, for example, Brown et al. (US Pat. No. 5,399,413 and US Pat. No. 5,499,623) and Vogelstein et al. (US Pat. No. 5,499,431, US Pat. No. 5,499,623 and US Pat. No. 5,499,633); the contents of each of which are incorporated herein by reference in their entirety. The distribution of target DNA molecules from the background among reactions follows Poisson statistics, and at the so-called "terminal dilution", the majority of reactions actually contain one or zero target DNA molecules. In another case, at the so-called "limiting dilution", some reactions contain zero DNA molecules, some contain one molecule, and often many molecules, following a Poisson distribution. It is understood that while terminal dilution and limiting dilution are useful concepts to describe the DNA loading amount into a reaction vessel, they do not have a formal mathematical definition and are not necessarily mutually exclusive. Ideally, the number of PCR positive reactions (PCR(+)) at the terminal dilution is equal to the number of template molecules originally present. Limiting dilution uses Poisson statistics to reveal the underlying amount of DNA. The advantage of the digital principle compared to qPCR is that it avoids the main underlying uncertainty of non-exponential amplification during the early cycles as well as any need to interpret the time dependence of the fluorescence intensity - the analog signal. [Prior art documents] [Patent documents]

[0007] [Patent Document 1] U.S. Pat. No. 6,143,496 [Patent Document 2] U.S. Patent No. 6,391,559 [Patent Document 3] U.S. Patent No. 6,440,706 [Patent Document 4] U.S. Patent No. 6,753,147 [Patent Document 5] U.S. Pat. No. 7,824,889 Summary of the Invention [Means for solving the problem]

[0008] The present invention generally relates to the manipulation of nucleic acids in droplets, and in particular to nucleic acid amplification and detection.In one embodiment, the present invention provides a droplet that contains a single nucleic acid template and multiple primer pairs specific to multiple target sites on the template.The single nucleic acid template can be DNA (e.g., genomic DNA, cDNA, etc.) or RNA.The template is amplified in the droplet for detection; and preferably, can be amplified using multiple primer pairs as described herein.

[0009] Being able to amplify and detect a single nucleic acid in a droplet allows digital PCR, detection, counting, and differentiation between nucleic acids, especially those present in a heterogeneous sample. Thus, the present invention is applied to digital amplification technology, and in certain embodiments, allows multiplex PCR in droplets. For example, multiplexing primers in droplets allows the number of PCR droplets to be increased while keeping the amount of input DNA the same or less, resulting in the same or greater amplicon yield. This results in an overall increase in the amount of PCR positive droplets and amplicon yield without consuming more DNA. Even if the number of PCR primer pairs per droplet is more than one, there is only one template molecule per droplet, and therefore in some implementations, only one primer pair per droplet will be utilized at a time. Thus, the advantage of droplet PCR in eliminating bias from allele-specific PCR or from competition between different amplicons is maintained. However, as described below with respect to the detection of haplotypes, other implementations advantageously allow the detection of multiple loci on a single template, preferably using multiple primer pairs designed to minimize bias.

[0010] The microfluidic droplets for multiplex analysis according to the present invention contain a plurality of probes that hybridize to the amplicons produced in the droplets. Preferably, the droplets contain more than one probe, for example, 2, 3, 4, 5, 6, 7, 8, 9, 10, 12, 14, 16, 18, 20, 25, 30, 35, 40, 45, 50, 55, 60, 65, 60, 75, 80, 85, 90, 95, 100, 110, 120, 130, 140, 150, 160, 170, 180, 190, 200, 500, or more probes. Certain members of the plurality of probes contain a detectable label. Each member of the plurality of probes may contain the same detectable label or may contain different detectable labels. The detectable labels are preferably fluorescent labels. The plurality of probes may include one or more groups of probes with different concentrations. The one or more groups of probes may include the same detectable label, and these probes will have different intensities when detected due to their different probe concentrations. The droplets of the present invention may further contain one or more reagents for performing a polymerase chain reaction, such as DNA or RNA polymerase and / or dNTPs.

[0011] The present invention further relates to a method for detecting multiple targets in a biological sample using digital PCR in a microfluidic droplet. The sample may be a human tissue or a body fluid. Exemplary body fluids: pus, sputum, semen, urine, blood, saliva, and cerebrospinal fluid.

[0012] Form one or more droplets, each of which contains a single nucleic acid template and a heterogeneous mixture of primer pairs and probes, each of which is specific to multiple target sites on the template.For example, a first fluid (continuous or, as in the case of droplets, discontinuous) containing a single nucleic acid template (DNA or RNA) is combined with a second fluid (also continuous or, as in the case of droplets, discontinuous) containing multiple primer pairs and multiple probes, each of which is specific to multiple target sites on the nucleic acid template, to form droplets containing a single nucleic acid template and a heterogeneous mixture of primer pairs and probes.The second fluid can also contain reagents for carrying out a PCR reaction, such as polymerase and dNTPs.

[0013] Certain members of the plurality of probes contain a detectable label. Each member of the plurality of probes may contain the same detectable label or may contain different detectable labels. The detectable label is preferably a fluorescent label. The plurality of probes may contain one or more groups of probes with different concentrations. One or more groups of probes may contain the same detectable label, and these probes vary in intensity upon detection due to their different probe concentrations.

[0014] The first and second fluids may each be in droplet form. Any technique known in the art for forming droplets may be used with the methods of the present invention. An exemplary method includes flowing a stream of sample fluid containing nucleic acid templates across two opposing streams of flowing carrier fluid. The carrier fluid is immiscible with the sample fluid. As a result of the crossing of the sample fluid with the two opposing streams of flowing carrier fluid, the sample fluid is compartmentalized into individual sample droplets containing the first fluid. The carrier fluid may be any fluid that is immiscible with the sample fluid. An exemplary carrier fluid is oil. In certain embodiments, the carrier fluid includes a surfactant, such as a fluorosurfactant. The same method may be used to create individual droplets from a second fluid containing primer pairs (and, in some implementations, amplification reagents). Either the droplet containing the first fluid or the droplet containing the second fluid, or both, may be formed and stored in a library for later merging, certain implementation aspects of which are described in U.S. Patent Application No. 12 / 504,764, which is incorporated herein by reference in its entirety for all purposes. Once formed, the droplet containing the first fluid and the droplet containing the second fluid can be merged to form a single droplet containing a single nucleic acid template and a heterogeneous mixture of primer pairs and probes. Merging can be accomplished, for example, in the presence of an electric field. Furthermore, it is not necessary that both fluids are in the form of droplets when merging. One exemplary method for merging fluid portions and droplets is taught, for example, in co-pending U.S. Patent Application No. 61 / 441,985, filed on the same day as this application.

[0015] The nucleic acid template in each of the combined / formed droplets is amplified, for example, by thermal cycling the droplets at a temperature / condition sufficient for carrying out a PCR reaction.The resulting amplicons in the droplets can then be analyzed.For example, the presence or absence of multiple targets in one or more droplets is optically detected, for example, by detectable labels on multiple probes.

[0016] The present invention further relates to a method for analyzing target nucleic acid. More specifically, the method of the present invention can detect polymerase errors occurring during PCR reaction, and can exclude from analysis amplification products that are the result of polymerase errors. The method of the present invention is particularly useful in digital PCR, where polymerase errors can result in a section of a sample that is erroneously identified as containing a mutant allele, i.e., a false positive. Such false positives generally have a large impact on the validity and accuracy of digital PCR results. The method of the present invention can uniquely detect multiple targets with the same light color. The method of the present invention is particularly useful in digital PCR, where it is desirable to identify multiple different target molecules that may be present in the starting test fluid.

[0017] The method of the present invention includes forming sample droplets containing target nucleic acids. Ideally, the method of the present invention includes forming droplets for digital PCR. A preferred digital PCR droplet contains one copy of the nucleic acid to be amplified, but may contain multiple copies of the same nucleic acid sequence. Any technique known in the art for forming sample droplets can be used with the method of the present invention. One exemplary method includes flowing a stream of sample fluid containing nucleic acids across two opposing streams of flowing carrier fluid. The carrier fluid is immiscible with the sample fluid. As a result of the crossing of the sample fluid with the two opposing streams of flowing carrier fluid, the sample fluid is compartmentalized into individual sample droplets. The carrier fluid can be any fluid that is immiscible with the sample fluid. An exemplary carrier fluid is an oil. In certain embodiments, the carrier fluid includes a surfactant, such as a fluorosurfactant.

[0018] The target is then amplified within the droplet. Any method known in the art can be used to amplify the target nucleic acid linearly or exponentially. A preferred method is polymerase chain reaction (PCR). For the present invention, any amplification technique generally known in the art can be performed, such as rolling circle amplification, isothermal amplification, or any combination of amplification methods using locus-specific primers, nested primers, or random primers (the aforementioned primers, and / or primers used in PCR, are included in the term "amplification reagent"). Once amplified, the droplets containing amplicons from the target and amplicons from variants of the target are eliminated. One method of eliminating droplets containing a heterogeneous population of amplicons from droplets containing a homogeneous population of amplicons includes hybridizing detectably labeled probes to the amplicons, flowing the droplets through a microfluidic channel, and eliminating droplets in which both amplicons from the target and amplicons from variants of the target are detected.

[0019] Once the droplets containing the heterogeneous population of amplicons are eliminated, the droplets containing the homogeneous population of amplicons are analyzed. Any analytical technique known in the art can be used. In certain embodiments, the analysis of the droplets includes determining the number of droplets containing only wild type target and determining the number of droplets containing only mutants of the target. In general, the presence of droplets containing only mutants is indicative of a disease, such as cancer. The mutants can be allelic variants. An exemplary allelic variant is a single nucleotide polymorphism. The mutants can also be specific haplotypes. Haplotypes refer to the presence of two or more variants on the same nucleic acid strand. Haplotypes can be more informative or predictive than genotypes when used to determine such things as the presence or severity of a disease, response to drug therapy, or drug resistance of a bacterial or viral infection. Because each droplet contains only one template strand, it is an ideal vessel for haplotype determination. Detection of two or more variants in a single droplet containing one intact nucleic acid strand identifies the haplotype of the variants on that strand. The presence of two or more markers in the same droplet can be identified by means of the presence of multiple dyes, or increased intensity of a single dye, or a combination of both. The haplotype of a sample can be determined by any method that allows for the identification of multiple variants in a single droplet.

[0020] According to some implementations of the invention, a method for analyzing target nucleic acids is provided, the method comprising compartmentalizing a first fluid into a plurality of portions, each portion containing a single target nucleic acid; amplifying the targets in the portions; eliminating portions containing amplicons from the targets and amplicons from variants of the targets; and analyzing the target amplicons.

[0021] In another aspect, the invention generally provides methods for detecting the recurrence of cancer in a patient. These methods may include forming sample droplets containing a single target nucleic acid taken from a patient sample, flowing the sample droplets through a channel, amplifying the target in the droplets, detecting the amplified target in the droplets, rejecting droplets containing a heterogeneous population of amplicons, and analyzing droplets that are not rejected to determine mutant alleles indicative of recurrence. In certain embodiments, the analyzing step includes capturing amplicons obtained from the droplets using a labeled capture probe. The sample may be human tissue or a bodily fluid. Exemplary bodily fluids are pus, sputum, semen, urine, blood, saliva, feces, and cerebrospinal fluid. In another aspect of the invention, the invention generally provides methods for forensically identifying low levels of a target nucleic acid in an environment with multiple other nucleic acid sources. Such methods may also be performed using fluids compartmented in a container other than or in addition to droplets. [Brief description of the drawings]

[0022] [Figure 1] FIG. 1 illustrates a droplet forming device. [Diagram 2] FIG. 2 illustrates a portion of the droplet forming device of FIG. [Diagram 3] Figure 3 illustrates an exemplary microfluidic system for droplet generation and readout: Figure 3a illustrates the liquid generation chip; Figure 3b illustrates droplet spacing for readout; and Figure 3c depicts a cartoon of droplet readout by fluorescence. [Figure 4]Figure 4 illustrates serial dilutions of template DNA quantified by dPCR. Figure 4a shows droplet fluorescence during readout for the highest concentration sample. Each distinct fluorescence burst corresponds to an individual droplet. Two distinct groups of droplets are clearly visible: PCR(+) droplets, which peak at ~0.8 V, and PCR(-) droplets, which peak at ~0.1 V; Figure 4b shows a histogram of peak fluorescence intensity of droplets from the full data trace in (a). PCR(+) and PCR(-) droplets appear as two very distinct populations centered at 0.78 and 0.10 V, respectively; Figure 4c shows serial dilutions of template DNA. Open circles: observed occupancy; solid line: best fit to Eqn2 (A=0.15, f=4.8, R2-0.9999). [Diagram 5] Figure 5A is a schematic diagram of a droplet having five sets of primers for PCR amplification of a template sequence and five probes, each labeled with a fluorescent dye and specifically binding to the amplified sequence; Figure 5B is a time trace of the fluorescence intensity detected from the droplet after PCR amplification; Figure 5C is a scatter plot showing clusters representing droplets containing specific amplified sequences (TERT, RNaseP, E1a, SMN1 and SMN2). [Figure 6] FIG. 6A is a schematic diagram of a droplet having five sets of primers for PCR amplification of a template sequence and five probes, each labeled with a fluorescent dye and specifically binding to the amplified sequence; FIG. 6B is a scatter plot showing clusters representing droplets containing specific amplified sequences (TERT, 815A, RNaseP, E1a, and 815G); FIG. 6C is a table showing the copy numbers of the specific sequences shown in FIG. 6B. [Figure 7] FIG. 7 is a schematic diagram illustrating single-color detection of one gene sequence on a microfluidic device. [Figure 8] FIG. 8 is a schematic diagram illustrating dual-color detection of two gene sequences in a microfluidic device. [Figure 9] FIG. 9 is a schematic diagram illustrating dual-color detection of three gene sequences in a microfluidic device. [Figure 10] FIG. 10 shows two dot plots illustrating clusters of gene sequences detected by fluorescence intensity. The left panel is a dot plot showing four clusters. Blocks of SMN1 sequence were present. Top left: microdroplets containing the reference sequence (SMARCC1); bottom left: microdroplets containing no sequence; bottom center: microdroplets containing the sequence of SMN1; and bottom right: microdroplets containing the sequence of SMN2. The right panel is a dot plot showing four clusters. Blocks of SMN1 sequence were not present. Top left: microdroplets containing the reference sequence (SMARCC1); bottom left: microdroplets containing no sequence; bottom center: microdroplets containing the sequence of SMN1; and bottom right: microdroplets containing the sequence of SMN2. The shift of the bottom center cluster in the right panel compared to the left panel confirms that fluorescence intensity provides a very sensitive measure of the presence of the sequence. [Figure 11] FIG. 11 illustrates histograms of duplex gene copy number assays using only one type of fluorophore by digital PCR; FIG. 11a illustrates histograms of droplet peak fluorescence intensity; FIG. 11b shows a comparison of gene copy numbers measured by single-color dPCR. [Figure 12] FIG. 12 is a schematic diagram of tuning the intensity of a detectable label to a specific target in a microfluidic device. [Figure 13] FIG. 13 is a line graph illustrating the linear dependence of droplet fluorescence intensity on probe concentration (straight line, best linear fit (y=-0.092x+0.082, R2=0.995). [Figure 14] Figure 14 illustrates a 5-plex dPCR assay for spinal muscular atrophy using only two fluorophores. Figure 14a is a 2D histogram of droplet fluorescence intensity shown as a heat map for a 5-plex assay on a synthetic model chromosome for validation. Six well-resolved droplet populations correspond to the five individual assays and a blank droplet; Figure 14b shows the results of an SMA pilot study. [Figure 15] Figure 15 illustrates a 9-plex dPCR assay for spinal muscular atrophy using only two fluorophores, showing the process of optimizing droplet intensity. Figures 15a and 15b are 2D histograms of droplet fluorescence intensity shown as heat maps for a 9-plex assay on a synthetic model chromosome, with warmer colors representing higher droplet numbers ((a)=before optimization; (b)=after optimization). [Figure 16] FIG. 16 illustrates an optical schematic for the combined use of optical labeling and multiplexing. [Figure 17] FIG. 17 illustrates a dPCR assay combining multiplexing and optical labeling using co-flow microfluidics. The contribution from all droplets, i.e. from three different triplex assays, is shown. (Both panels) 2-D histograms shown as heat maps, with warmer colors representing higher droplet numbers. (Left panel) Optical label histogram, i.e., droplet fluorescence intensity measured at wavelengths for the two fluorophores that comprise the optical label. (Right panel) Assay histogram, i.e., droplet fluorescence intensity measured at wavelengths suitable for FAM detection (x-axis) and VIC detection (y-axis). Both histograms were corrected for spectral overlap by standard techniques. [Figure 18] Figure 18 shows single assay selection using optical labels. Selections were made from all of the droplets from Figure 17. Each of the three different selections in panels A-C was for optical labels encoding the same assay (TERT, SMN1, and SMN2). Histograms are as described in Figure 17. (Left histogram, optical labels) Superimposed lines define the bounding box boundaries for selecting single optical labels. (Right histogram, assays) Only droplets containing the selected optical labels are displayed. [Figure 19]FIG. 19 shows single assay selection using optical labels. Selections were made from all of the droplets from FIG. 17. Each of the three different selections in panels A-C was for optical labels encoding the same assay (TERT, c.5C from SMN1, and BCKDHA). Histograms are as described in FIG. 17. (Left histogram, optical labels) Superimposed lines define the bounding box boundaries for selecting single optical labels. (Right histogram, assays) Only droplets containing the selected optical labels are displayed. [Figure 20] FIG. 20 shows single assay selection using optical labels. Selections were made from all of the droplets from FIG. 17. Each of the three different selections in panels A-C was for optical labels encoding the same assay (TERT, c.88G from SMN1, and RNaseP). Histograms are as described in FIG. 17. (Left histogram, optical labels) Superimposed lines define the bounding box boundaries for selecting single optical labels. (Right histogram, assays) Only droplets containing the selected optical labels are displayed. [Figure 21] FIG. 21 illustrates a combined multiplexed and optically labeled dPCR assay using droplet merging. [Figure 22] FIG. 22 is a schematic diagram showing haplotype detection within droplets. DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS

[0023] The present invention provides materials and methods for the analysis of biomolecules.In one embodiment, the present invention provides for digital analysis in droplets, such as microfluidic droplets.The present invention allows digital PCR to be performed, and provides for the significant reduction or elimination of errors.

[0024] Ideally, the sensitivity of digital PCR is limited only by the number of independent amplifications that can be analyzed, which has motivated the development of several ultra-high throughput miniaturized methods that can perform millions of single molecule PCR reactions in parallel (discussed in detail elsewhere). In a preferred embodiment of the present invention, digital PCR is performed in aqueous droplets separated by oil using a microfluidics system. In another preferred embodiment, the oil is a fluorinated oil, such as Fluorinert oil (3M). In an even more preferred embodiment, the fluorinated oil contains a surfactant, such as a PFPE-PEG-PFPE triblock copolymer, to stabilize the droplets from coalescing during the amplification step or at any point when they come into contact with each other. The microfluidic approach involves the use of a large number of (e.g., 10 μm) microfluidic devices that act as picoliter volume reaction vessels. 6 This allows for the rapid generation of droplets of very uniform size (see review of droplet-based microfluidics). However, as will be explained, the invention is not limited to dPCR performed in water-in-oil emulsions, but rather is general to all reaction compartmentalization methods for dPCR. In the description that follows, the invention is described with respect to the use of droplets for compartmentalization, but it is considered that this choice of description is not limiting for the invention and that the method of the invention is compatible with all other reaction compartmentalization methods for dPCR.

[0025] The method of the present invention involves a novel strategy for the simultaneous performance of multiple different amplification reactions on the same sample to quantify the abundance of multiple different DNA targets, commonly known to those skilled in the art as "multiplexing". The method of the present invention for multiplexed dPCR assays promises greater plexity - the number of simultaneous reactions - than is possible with existing qPCR or dPCR techniques. It is based on the differential nature of amplification at the end point or limiting dilution, which occurs most often because only a single target allele is ever present in any one droplet, even when multiple primers / probes targeting different alleles are present. This alleviates the complications that would otherwise result from competing reactions occurring simultaneously, such as different times to reach exponential phase and unintended reactions between primers.

[0026] In one embodiment, the present invention provides materials and methods for improving amplicon yield while maintaining sensitivity and specificity in droplet-based digital PCR.More specifically, the present invention provides the droplet that contains a single nucleic acid template and multiplex PCR primer, and provides the method for detecting multiple targets in biological samples by forming such droplet and amplifying nucleic acid template using droplet-based digital PCR.

[0027] Reactions in microfluidic droplets produce a very uniform fluorescence intensity at the end point, which ultimately depends on the probe hydrolysis efficiency. Thus, in another embodiment of the method of the present invention, different reactions with different efficiencies can be distinguished based on the end point fluorescence intensity alone, even if they have the same color. Furthermore, in another method of the present invention, the efficiency can be tuned by simply adjusting the probe concentration, resulting in a versatile multiplexing method that is easy to use. In one demonstration of the present invention, a 5-plex TaqMan® dPCR assay worked "right out of the box," as opposed to the tedious optimization that typifies qPCR that is multiplexed to this extent. In another embodiment of the present invention, the addition of multiple colors increases the number of possible reactions exponentially, rather than linearly as in qPCR, since individual reactions can be labeled with multiple fluorophores. As an example, two fluorophores (VIC and FAM) were used to distinguish five different reactions in a single run of the present invention.

[0028] The method of the present invention can detect polymerase errors that occur during the amplification reaction and can exclude products that are the result of polymerase errors from the analysis. Essentially, the method of the present invention increases the sensitivity of digital PCR by identifying amplification products that are false positives and excluding them from the analysis.

[0029] The methods of the invention include forming sample droplets containing a single target nucleic acid, amplifying the target within the droplets, rejecting droplets containing amplicons from the target and amplicons from variants of the target, and analyzing the target amplicons.

[0030] Nucleic Acid Target Molecules Nucleic acid molecules include deoxyribonucleic acid (DNA) and / or ribonucleic acid (RNA). Nucleic acid molecules can be synthesized or obtained from naturally occurring sources. In one embodiment, the nucleic acid molecules are isolated from a biological sample that contains various other components, such as proteins, lipids, and non-template nucleic acids. The nucleic acid template molecule can be obtained from any cellular material obtained from an animal, plant, bacteria, fungus, or any other cellular organism. In certain embodiments, the nucleic acid molecule is obtained from a single cell. Biological samples for use in the present invention include virus particles or preparations. Nucleic acid molecules can be obtained directly from an organism or from biological samples obtained from an organism, such as blood, urine, cerebrospinal fluid, semen, saliva, sputum, feces, and tissues. Any tissue or body fluid specimen can be used as a source of nucleic acid for use in the present invention. Nucleic acids can also be isolated from cultured cells, such as primary cell cultures or cell lines. The cells or tissues from which the template nucleic acid is obtained may be infected with a virus or other intracellular pathogen. The sample may be total RNA extracted from a biological specimen, a cDNA library, a virus or genomic DNA. In certain embodiments, the nucleic acid molecule is determined relative to another target molecule, such as a protein, an enzyme, a substrate, an antibody, a binder, a bead, a small molecule, a peptide or any other molecule, and serves as a surrogate for quantifying and / or detecting the target molecule.

[0031] In general, nucleic acids can be extracted from biological samples by a variety of techniques, such as those described by Maniatis et al., Molecular Cloning: A Laboratory Manual, Cold Spring Harbor, NY, pp. 280-281 (1982). Nucleic acid molecules can be single-stranded, double-stranded, or double-stranded with single-stranded regions (e.g., stem- and loop-structures).

[0032] droplet formation The method of the present invention includes forming sample droplets, some of which contain zero target nucleic acid molecules, some of which contain one target nucleic acid molecule, and some of which may or may not contain multiple nucleic acid molecules (corresponding to limit or end-point dilution, respectively, as defined above). In a preferred embodiment, the distribution of molecules in the droplets follows a Poisson distribution. However, methods for non-Poisson loading of droplets are known to those skilled in the art, including, but not limited to, active sorting of droplets, such as by laser-induced fluorescence, or by passive one-to-one loading. Although the following description assumes Poisson loading of droplets, such description is not intended to exclude non-Poisson loading, as the present invention is compatible with all distributions of DNA loading amounts that comply with limit or end-point dilution.

[0033] The droplets are aqueous droplets surrounded by an immiscible carrier fluid.Methods for forming such droplets are shown, for example, in Link et al. (US Patent Application Nos. 2008 / 0014589, 2008 / 0003142, and 2010 / 0137163), Stone et al. (US Patent No. 7,708,949 and US Patent Application No. 2010 / 0172803), Anderson et al. (US Patent No. 7,041,481 and reissued as Reissue No. 41,780) and Raindance Technologies Inc., European Patent Application Publication No. 2047910.The contents of each of these are incorporated herein by reference in their entirety.

[0034] FIG. 1 shows an exemplary embodiment of a device 100 for droplet formation. The device 100 comprises an inlet channel 101, an outlet channel 102, and two carrier fluid channels 103 and 104. The channels 101, 102, 103, and 104 meet at a junction 105. The inlet channel 101 flows a sample fluid to the junction 105. The carrier fluid channels 103 and 104 flow a carrier fluid that is immiscible with the sample fluid to the junction 105. The inlet channel 101 narrows at its tip where it connects to the junction 105 (see FIG. 2). The inlet channel 101 is oriented such that it is perpendicular to the carrier fluid channels 103 and 104. Droplets are formed when the sample fluid flows from the inlet channel 101 to the junction 105 where it intersects with the flowing carrier fluid that is supplied to the junction 105 by the carrier fluid channels 103 and 104. The outlet channel 102 receives droplets of sample fluid surrounded by carrier fluid.

[0035] The sample fluid is generally an aqueous buffer solution, such as ultrapure water (e.g., obtained by column chromatography, e.g., resistivity of 18 megaohms), 10 mM Tris HCl and 1 mM EDTA (TE) buffer, phosphate buffered saline (PBS) or acetate buffer. Any liquid or buffer that is physiologically compatible with nucleic acid molecules can be used. The carrier fluid is one that is immiscible with the sample fluid. The carrier fluid can be a non-polar solvent, decane (e.g., tetradecane or hexadecane), fluorocarbon oil, silicone oil or other oil (e.g., mineral oil).

[0036] In certain embodiments, the carrier fluid contains one or more additives, such as agents that increase, reduce or otherwise create non-Newtonian surface tension (surfactants) and / or agents that stabilize droplets from spontaneously coalescing upon contact. Surfactants can include Tween, Span, fluorosurfactants, and other agents that are more soluble in oil than in water. In some applications, performance is improved by adding a second surfactant or other agent, such as a polymer or other additive, to the sample fluid. Surfactants can help control or optimize droplet size, flow rate, and uniformity, for example, by reducing the shear force required to push or inject droplets into the crossing channel. This can affect droplet volume and periodicity, or the rate or frequency at which droplets break off into the crossing channel. Additionally, surfactants can help stabilize aqueous emulsions in fluorinated oil from coalescing.

[0037] In certain embodiments, the droplets can be coated with a surfactant or a mixture of surfactants.Preferred surfactants that can be added to the carrier fluid include, but are not limited to, surfactants such as sorbitan-based carboxylic acid esters (e.g., "Span" surfactants, Fluka Chemika), including sorbitan monolaurate (Span 20), sorbitan monopalmitate (Span 40), sorbitan monostearate (Span 60) and sorbitan monooleate (Span 80) and perfluorinated polyethers (e.g., DuPont Krytox 157 FSL, FSM and / or FSH). Other non-limiting examples of nonionic surfactants that can be used include polyoxyethylenated alkylphenols (e.g., nonyl-, p-dodecyl-, and dinonylphenol), polyoxyethylenated straight chain alcohols, polyoxyethylenated polyoxypropylene glycols, polyoxyethylenated mercaptans, long chain carboxylic acid esters (e.g., glyceryl and polyglyceryl esters of natural fatty acids, propylene glycol, sorbitol, polyoxyethylenated sorbitol esters, polyoxyethylene glycol esters, etc.), and alkanolamines (e.g., diethanolamine-fatty acid condensates and isopropanolamine-fatty acid condensates).

[0038] In certain embodiments, the carrier fluid can be made to flow through the outflow channel such that the surfactant in the carrier fluid coats the channel walls. In one embodiment, the fluorosurfactant can be prepared by reacting perfluorinated polyether DuPont Krytox 157 FSL, FSM or FSH with aqueous ammonium hydroxide in a volatile fluorinated solvent. The solvent and residual water and ammonia can be removed with a rotary evaporator. The surfactant can then be dissolved (e.g., 2.5 wt%) in fluorinated oil (e.g., Flourinert (3M)), which then serves as the carrier liquid.

[0039] One approach to combining sample fluids using a device called a lambda injector involves forming droplets and contacting the droplets with a fluid stream, where a portion of the fluid stream merges with the droplets to form a mixed droplet. In this approach, only one phase needs to reach the merge area within a droplet of one form. Further description of such methods is provided in commonly owned and co-pending U.S. patent application to Yurkovetsky et al. (U.S. Patent Application No. 61 / 441,985), the contents of which are incorporated herein by reference in their entirety.

[0040] According to the method of operation of the lambda injector, droplets are formed as described above. After the formation of the sample droplets from the first sample fluid, the droplets are contacted with a flow of a second sample fluid stream. As a result of the contact of the droplets with the fluid stream, a portion of the fluid stream combines with the droplets to form a mixed droplet.

[0041] Droplets of a first sample fluid separated from one another by and suspended in an immiscible carrier fluid flow through a first channel. The droplets are delivered to a merge area, i.e., the junction of the first and second channels, by pressure-driven flow generated by a positive displacement pump. When the droplets reach the merge area, a bolus of the second sample fluid is forced out of the opening of the second channel into the first channel. Preferably, the channels are oriented perpendicular to one another. However, any angle that results in the channels being orthogonal can be used.

[0042] The bolus of the second sample fluid stream continues to increase in size due to the pumping action of a positive displacement pump connected to the channel, which produces a steady stream of the second sample fluid into the merge area. The flowing droplets containing the first sample fluid eventually come into contact with the bolus of the second sample fluid being pushed into the first channel. As a result of the contact between the two sample fluids, a portion of the second sample fluid splits off from the second sample fluid stream and merges with the first sample fluid droplets to form mixed droplets. In certain embodiments, each incoming droplet of the first sample fluid is merged with an equal amount of the second sample fluid.

[0043] In certain embodiments, the first and second sample fluids are charged. Descriptions of charging sample fluids are provided in Link et al. (US Patent Application No. 2007 / 0003442) and Raindance Technologies Inc. European Patent No. 2004316, the contents of each of which are incorporated herein by reference in their entirety. Any suitable technique can be used to create charge in the first and second sample fluids in the carrier fluid, for example, by placing the first and second sample fluids in an electric field (which can be AC, DC, etc.) and / or by causing a reaction, such as a chemical reaction, an ionic reaction, a photocatalytic reaction, etc., that causes the first and second sample fluids to have a charge.

[0044] In some embodiments, the electric field is generated from an electric field generator, i.e., a device or system capable of creating an electric field that can be applied to a fluid. The electric field generator can generate an AC field (i.e., one that varies periodically with time, e.g., sinusoidally, sawtooth, square, etc.), a DC field (i.e., one that is constant with time), a pulsed field, etc. The electric field generator can be constructed and arranged to create an electric field in the fluid contained within the channel or microfluidic channel. According to some embodiments, the electric field generator can be integral with or separate from the fluid system that includes the channel or microfluidic channel.

[0045] Techniques for generating a suitable electric field (which may be AC, DC, etc.) are known to those skilled in the art. For example, in one embodiment, the electric field is generated by applying a voltage across a pair of electrodes, which may be disposed on or embedded within the fluid system (e.g., in a substrate defining a channel or microfluidic channel) and / or disposed proximate to the fluid such that at least a portion of the electric field interacts with the fluid. The electrodes may be made of any suitable electrode material or materials known to those skilled in the art, including (but not limited to) silver, gold, copper, carbon, platinum, copper, tungsten, tin, cadmium, nickel, indium tin oxide ("ITO"), etc., and combinations thereof. In some cases, transparent or substantially transparent electrodes may be used.

[0046] The electric field encourages the rupture of the interface separating the droplet from the second sample fluid. The rupture of the interface encourages the merging of the first sample fluid droplet with the bolus of the second sample fluid. The formed mixed droplet continues to grow until a portion of the second sample fluid is released or splits off from the second sample fluid stream before the arrival and merging of the next droplet containing the first sample fluid. Splitting of said portion of the second sample fluid from the second sample fluid stream occurs as soon as the shear force exerted by the miscible carrier fluid on the formed mixed droplet overcomes the surface tension forces (whose effect is to keep the split off portion of the second sample fluid connected to the second sample fluid stream). The now fully formed mixed droplet continues to flow through the first channel.

[0047] In other embodiments, the rupture at the interface may be spontaneous or the rupture may be assisted by surface chemistry. The invention is not limited with respect to the method of rupture at the interface, as the rupture may be caused by any means.

[0048] In the context of PCR, in a preferred embodiment, the first sample fluid contains a nucleic acid template. Droplets of the first sample fluid are formed as described above. These droplets will contain a nucleic acid template. In certain embodiments, the droplets will contain only a single nucleic acid template, and therefore digital PCR can be performed. The second sample fluid contains reagents for the PCR reaction. Such reagents generally include Taq polymerase, deoxynucleotides of types A, C, G and T, magnesium chloride, and forward and reverse primers, all suspended in an aqueous buffer. The second fluid also contains a detectably labeled probe for detecting the amplified target nucleic acid, the details of which are discussed below. The nucleic acid-containing droplets are then combined with PCR reagents in the second fluid as described above to produce droplets containing Taq polymerase, deoxynucleotides of types A, C, G and T, magnesium chloride, forward and reverse primers, detectably labeled probes, and target nucleic acid. In another embodiment, the first fluid can contain template DNA and PCR master mix (defined below), and the second fluid can contain forward and reverse primers and probes. The present invention is not limited in any way with respect to the configuration of the first and second fluidics components for PCR or digital PCR. For example, in some embodiments, the template DNA is contained in the second fluid within a droplet.

[0049] Target Amplification The method of the present invention further includes amplifying the target nucleic acid in each droplet. Amplification refers to the production of additional copies of a nucleic acid sequence and is generally performed using polymerase chain reaction or other techniques known in the art (e.g., Dieffenbach and Dveksler, PCR Primer, a Laboratory Manual, Cold Spring Harbor Press, Plainview, NY

[1995] ). The amplification reaction can be any amplification reaction known in the art that amplifies nucleic acid molecules, such as polymerase chain reaction, nested polymerase chain reaction, ligase chain reaction (Barany F. (1991) PNAS 88:189-193; Barany F. (1991) PCR Methods and Applications 1:5-16), ligase detection reaction (Barany F. (1991) PNAS 88:189-193), strand displacement amplification, transcription-based amplification system, nucleic acid sequence-based amplification, rolling circle amplification and hyperbranched rolling circle amplification.

[0050] In certain embodiments, the amplification reaction is a polymerase chain reaction.Polymerase chain reaction (PCR) refers to a method by KB Mullis (U.S. Pat. Nos. 4,683,195 and 4,683,202, which are incorporated herein by reference) for increasing the concentration of a segment of a target sequence in a mixture of genomic DNA without cloning or purification.The process for amplifying a target sequence involves introducing excess oligonucleotide primers into a DNA mixture containing the desired target sequence, followed by a precise sequence of thermal cycling in the presence of DNA polymerase.The primers are complementary to their respective strands of a double-stranded target sequence.

[0051] To achieve amplification, primers are annealed to their complementary sequences in the target molecule. After annealing, the primers are extended with polymerase to form new complementary strand pairs. The steps of denaturation, primer annealing and polymerase extension can be repeated many times (i.e., denaturation, annealing and extension constitute one cycle; there can be many cycles), resulting in a highly concentrated amplified segment of the desired target sequence. The length of the amplified segment of the desired target sequence is determined by the relative positions of the primers to each other and by the cycling parameters, and therefore this length is a controllable parameter.

[0052] Methods for performing PCR in droplets are provided, for example, in Link et al. (U.S. Patent Application Nos. 2008 / 0014589, 2008 / 0003142, and 2010 / 0137163), Anderson et al. (U.S. Patent No. 7,041,481 and reissued as Reissue No. 41,780), and Raindance Technologies Inc., European Patent Application Publication No. 2047910, the contents of each of which are incorporated herein by reference in their entireties.

[0053] Sample droplets may be premixed with primer(s) or primer(s) may be added to the droplets. In some embodiments, droplets made by splitting the starting sample are combined with a second set of droplets containing one or more primers for target nucleic acid to produce final droplets. Droplet merging can be accomplished using one or more droplet merging techniques described, for example, in Link et al. (U.S. Patent Application Nos. 2008 / 0014589, 2008 / 0003142, and 2010 / 0137163) and Raindance Technologies Inc.'s European Patent Application Publication No. 2047910.

[0054] In embodiments involving droplet merging, two droplet formation modules are used. In one embodiment, the first droplet formation module produces sample droplets that correspond to the limit or end-point dilution of the target nucleic acid. The second droplet formation or reinjection module inserts droplets that contain reagents for a PCR reaction. Such droplets generally contain a "PCR master mix" (known to those skilled in the art as a mixture containing at least Taq polymerase, deoxynucleotides of types A, C, G and T, and magnesium chloride) and forward and reverse primers (collectively known to those skilled in the art as "primers"), all suspended in an aqueous buffer. The second droplet also contains a detectably labeled probe for detecting the amplified target nucleic acid, the details of which are discussed below. Different combinations of reagents between the two droplet types are contemplated. For example, in another embodiment, the template droplet also contains a PCR master mix, but the primers and probes remain in the second droplet. Any combination of reagents and template DNA can be used in accordance with the present invention.

[0055] Primers can be prepared by a variety of methods, including, but not limited to, cloning appropriate sequences and direct chemical synthesis using methods well known in the art (Narang et al., Methods Enzymol., 68:90 (1979); Brown et al., Methods Enzymol., 68:109 (1979)). Primers can also be obtained from commercial sources, such as Operon Technologies, Amersham Pharmacia Biotech, Sigma and Life Technologies. The primers can have the same melting temperature. The length of the primers can be extended or shortened at the 5' or 3' end to generate primers with desired melting temperatures. Also, the annealing position of each primer pair can be designed so that the sequence and length of the primer pair will produce the desired melting temperature. The simplest equation for determining the melting temperature of a primer smaller than 25 base pairs is the Wallace Rule (Td=2(A+T)+4(G+C)). Another method for determining the melting temperature of primers is the nearest neighbor method. Primers can also be designed using computer programs, including but not limited to Array Designer Software (Arrayit Inc.), Oligonucleotide Probe Sequence Design Software for Genetic Analysis (Olympus Optical Co.), NetPrimer, and DNAsis from Hitachi Software Engineering. The TM (melting or annealing temperature) of each primer is calculated using software programs such as Oligo Design available from Invitrogen Corp.

[0056] In one embodiment, the droplet formation module is arranged and controlled to cause interdigitation of sample droplets and PCR reagent droplets flowing through the channel. Such an arrangement is described, for example, in Link et al. (U.S. Patent Applications 2008 / 0014589, 2008 / 0003142, and 2010 / 0137163) and in Raindance Technologies Inc.'s European Patent Application Publication No. 2047910.

[0057] The sample droplets are then combined with the PCR reagent droplets to generate droplets containing the PCR master mix, primers, detectably labeled probes, and target nucleic acid. For example, droplets can be combined by using an electric field gradient to generate a dielectrophoretic force on the droplets and control the force to cause the droplets to merge; by generating droplets of different sizes and therefore moving at different speeds, thereby causing the droplets to merge; and by generating droplets of different viscosities and therefore moving at different speeds, thereby causing the droplets to merge with each other. Each of these techniques is further described in Link et al. (U.S. Patent Application Nos. 2008 / 0014589, 2008 / 0003142, and 2010 / 0137163) and in Raindance Technologies Inc.'s European Patent Application Publication No. 2047910. Further description of the generation and control of dielectrophoretic forces on droplets to cause them to coalesce is provided in Link et al. (US Patent Application Publication No. 2007 / 0003442) and in Raindance Technologies Inc.'s European Patent Application Publication No. 2004316.

[0058] In another embodiment, called simple droplet generation, a single droplet formation module or multiple droplet formation modules are arranged to generate droplets from a mixture that already contains template DNA, PCR master mix, primers, and detectably labeled probes. In yet another embodiment, called co-flow, two channels cross upstream of a single droplet formation module, thereby converging two flow streams. One flow stream contains one set of reagents and template DNA, and the other contains the remaining reagents. In a preferred embodiment for co-flow, the template DNA and PCR master mix are in one flow stream, and the primers and probes are in the other. However, the invention is not limited with respect to the composition of the elements of either flow stream. For example, in another embodiment, one flow stream contains only template DNA, and the other contains PCR master mix, primers, and probes. When the flow streams converge at a fluid intersection, they may or may not mix before the droplet generation nozzle. In either embodiment, some amount of fluid from the first stream and some amount of fluid from the second stream are encapsulated within a single droplet. Complete mixing occurs after encapsulation.

[0059] After the final droplets are generated by any of the droplet forming embodiments above or by any other embodiment, the droplets are subjected to thermal cycling to cause the amplification of the target nucleic acid in each droplet.In certain embodiments, the droplets are collected off-chip as emulsion in a PCR thermal cycling tube, and then subjected to thermal cycling in a conventional thermal cycler.The temperature profile for thermal cycling can be adjusted and optimized in the same way as any conventional DNA amplification by PCR.

[0060] In certain embodiments, the nucleic acid in the droplets is amplified by flowing the droplets through a channel in a serpentine path between heating and cooling lines. The width and depth of the channel may be adjusted to set the residence time at each temperature, and the residence time can be controlled anywhere between subseconds and minutes.

[0061] In certain embodiments, three temperature zones are used in the amplification reaction.The three temperature zones are controlled to cause double-stranded nucleic acid denaturation (high temperature zone), primer annealing (low temperature zone) and single-stranded nucleic acid amplification to produce double-stranded nucleic acid (middle temperature zone).The temperatures in these zones are within the ranges known in the art for carrying out PCR reaction.See, for example, Sambrook et al. (Molecular Cloning, A Laboratory Manual, 3rd Edition, Cold Spring Harbor Laboratory Press, Cold Spring Harbor, New York, 2001).

[0062] In certain embodiments, the three temperature zones are controlled to have temperatures as follows: 95° C. (T H ), 55℃(T L ), 72℃(T M The prepared sample droplets flow through the channel at a controlled rate. The sample droplets are placed in an initial denaturation zone (T H) first. The initial preheat is an extension zone where the nucleic acids in the sample droplets are successfully denatured prior to thermal cycling. The preheat zone requirements and the length of denaturation time required depend on the chemistry that will be used in the reaction. The sample passes through a high temperature zone of approximately 95°C where the sample is first separated into single stranded DNA in a process called denaturation. The sample then flows to a low temperature of approximately 55°C where the hybridization process occurs during which the primers anneal to the complementary sequences of the sample. Finally, the sample flows through a third intermediate temperature of approximately 72°C where the polymerase process occurs during which the primers are extended along the single strands of DNA with a thermostable enzyme. Methods for controlling the temperature in each zone can include, but are not limited to, electrical resistance, Peltier junctions, microwave irradiation, and illumination with infrared light.

[0063] As the nucleic acid flows through the channel, it undergoes the same thermal cycling and chemical reactions as the droplets go through each thermal cycle. The total number of cycles in the device is easily modified by expanding the thermal zones or by creating a continuous loop structure. The sample undergoes the same thermal cycling and chemical reactions as it would go through N amplification cycles in a full thermal device.

[0064] In other embodiments, the temperature zones are controlled to provide two separate temperature zones for the PCR reaction. In certain embodiments, the two temperature zones are controlled to have temperatures as follows: 95° C. (T H ) and 60°C (T L ). The sample droplets optionally flow through an initial preheat zone before entering thermal cycling. This preheat zone may be important for some chemistry for activation and also to ensure sufficient denaturation of the double-stranded nucleic acid within the droplet before the thermal cycling reaction begins. In an exemplary embodiment, the preheat dwell length provides approximately 10 minutes of preheating of the droplet at a higher temperature.

[0065] The sample droplet continues to a hot zone of approximately 95°C where the sample is first separated into single stranded DNA in a process called denaturation. The sample then flows through the device to a cold zone of approximately 60°C where a hybridization process occurs during which primers anneal to complementary sequences in the sample. Finally, a polymerase process occurs during which the primers are extended along the single strands of DNA with a thermostable enzyme. The sample undergoes the same thermal cycling and chemical reactions as it would through each thermal cycle of the complete device. The total number of cycles in the device is easily modified by block length and piping.

[0066] In another embodiment, droplets are created and / or merged on-chip and then stored on the same or another chip or off-chip in some type of storage container, such as a PCR tube, and the entire chip or storage container containing the droplets is then cycled to achieve the desired PCR heating and cooling cycles.

[0067] In another embodiment, the droplets are collected in a chamber where the oil can be rapidly exchanged due to the density difference between the droplet and the surrounding oil without removing the droplet. The temperature of the droplets can then be easily changed by exchanging the oil in the container with oil of a different temperature. This technique is generally useful for two and three step temperature cycling or any other temperature sequence.

[0068] The present invention is not limited by the method by which the droplets are thermally cycled. Any droplet thermal cycling method can be used.

[0069] Target Detection After amplification, the droplets are flowed to a detection module for detection of the amplified products. For embodiments in which the droplets are thermally cycled off-chip, the droplets require reinjection into a second fluidic circuit, which may or may not be on the same chip as the fluidic circuit(s) for droplet generation, for readout, or in certain embodiments, the droplets may be reinjected back into the original fluidic circuit used for droplet generation for readout. The droplets can be individually analyzed and detected using any method known in the art, such as detecting the presence or amount of reporters. In general, the detection module is in communication with one or more detection devices. The detection devices can be optical or electrical detectors or combinations thereof. Examples of suitable detection devices include optical waveguides, microscopes, diodes, photostimulation devices (e.g., lasers), photomultipliers and processors (e.g., computers and software), and combinations thereof that work together to detect the signal indication of a property, marker or reporter and to determine and command a measurement or sorting operation in a sorting module. Further description of detection modules and methods for detecting amplification products in droplets is provided in Link et al. (U.S. Patent Applications Nos. 2008 / 0014589, 2008 / 0003142, and 2010 / 0137163) and in European Patent Application Publication No. 2047910 to Raindance Technologies Inc.

[0070] In certain embodiments, the amplified target is detected using a detectably labeled probe. In certain embodiments, the detectably labeled probe is an optically labeled probe, such as a fluorescently labeled probe. Examples of fluorescent labels include Atto dye, 4-acetamido-4'-isothiocyanatostilbene-2,2'disulfonic acid; acridine and derivatives; acridine, acridine isothiocyanate; 5-(2'-aminoethyl)aminonaphthalene-1-sulfonic acid (EDANS); 4-amino-N-[3-vinylsulfonyl)phenyl]naphthalimide-3,5 disulfonate; N-(4-anilino-1-naphthyl)maleimide; anthranilamide; BODIPY; Brilliant Yellow. ;Coumarins and derivatives;Coumarin, 7-amino-4-methylcoumarin (AMC, coumarin 120), 7-amino-4-trifluoromethylcouluarin (coumaran 151);Cyanine dyes;Cyanosine;4',6-diamidino-2-phenylindole (DAPI);5',5''dibromopyrogallol-sulfonaphthalein (bromopyrogallol red);7-diethylamino-3-(4'- isothiocyanatophenyl)-4-methylcoumarin;diethylenetriaminepentaacetate;4,4'-diisothiocyanatodihydro-stilbene-2,2'-disulfonic acid;4,4'-diisothiocyanatostilbene-2,2'-disulfonic acid;5-[dimethylamino]naphthalene-1-sulfonyl chloride (DNS, dansyl chloride);4-dimethylaminophenylazophenyl-4'-isothiocyanate (DABITC);Eosin and derivatives;Eosin, eosin isothioate cyanate, erythrosine and derivatives;erythrosine B, erythrosine, isothiocyanate;ethidium;fluorescein and derivatives;5-carboxyfluorescein (FAM), 5-(4,6-dichlorotriazin-2-yl)aminofluorescein (DTAF), 2',7'-dimethoxy-4'5'-dichloro-6-carboxyfluorescein, fluorescein, fluorescein isothiocyanate, QFITC, (XRITC);fluorescamine;IR144;IR1446;Malachite Green Isothiocyanate;4-Methylumbelliferone Orthocresolphthalein;Nitrotyrosine;Pararosaniline;Phenol Red;B-Phycoerythrin;o-Phthaldialdehyde;Pyrene and derivatives;Pyrene, Pyrene Butyrate, Succinimidyl 1-Pyrene;Butyrate Quantum Dots;Reactive Red 4 (Cibacron Brilliant Red) 3B-A) Rhodamine and derivatives; including, but not limited to, 6-carboxy-X-rhodamine (ROX), 6-carboxyrhodamine (R6G), Lissamine rhodamine B sulfonyl chloride rhodamine (Rhod), rhodamine B, rhodamine 123, rhodamine X isothiocyanate, sulforhodamine B, sulforhodamine 101, sulfonyl chloride derivative of sulforhodamine 101 (Texas Red); N,N,N',N' tetramethyl-6-carboxyrhodamine (TAMRA); tetramethylrhodamine; tetramethylrhodamine isothiocyanate (TRITC); riboflavin; rosolic acid; terbium chelate derivatives; Cy3; Cy5; Cy5.5; Cy7; IRD 700; IRD 800; La Jolla Blue; phthalocyanines; and naphthalocyanines. Preferred fluorescent labels are FAM and VIC™ (from Applied Biosystems). Labels other than fluorescent labels are contemplated by the present invention, including other optically detectable labels.

[0071] In certain embodiments, the droplets of the present invention contain multiple detectable probes that hybridize to the amplicons produced within the droplet. Each member of the plurality of probes may contain the same detectable label, or may contain different detectable labels. The plurality of probes may also include one or more groups of probes at different concentrations. The groups of probes at different concentrations may contain the same detectable label, which varies in intensity due to the different probe concentrations.

[0072] In another embodiment, detection can be performed by scanning droplets confined to a monolayer in a storage device that is transparent to the method or wavelength for detection. Droplets stored in this manner can be scanned either by moving the storage device through a scanner or by moving the scanner over the storage device.

[0073] The present invention is not limited to the TaqMan assay as described above, but rather the invention encompasses the use of all fluorogenic DNA hybridization probes, such as molecular beacons, Solaris probes, Scorpion probes, and any other probes that function by sequence-specific recognition of target DNA by hybridization and result in an increase in fluorescence upon amplification of that target sequence.

[0074] Digital PCR performance in droplets Digital PCR performance in emulsion format was verified by measuring serial dilutions of a reference gene, branched-chain ketoacid dehydrogenase E1 (BCKDHA). Using a droplet generation microfluidic chip, a mixture of PCR master mix, 1× primers, probe for BCKDHA, and various concentrations of human genomic DNA mixture (1:1 NA14091 and NA13705) was compartmentalized into over one million 5.3 pL droplets in a fluorinated water-in-oil emulsion. The emulsion was subjected to thermal cycling off-chip, after which the fluorescence of each droplet was analyzed by fluorescence on the readout chip (see Figure 3).

[0075] An exemplary microfluidic system for droplet generation and readout is illustrated in FIG. 3. Microfluidic system for droplet generation and readout. As shown in FIG. 3a (droplet generation chip), a continuous aqueous phase containing PCR master mix, primers, probes, and template DNA flowed from the left into the fluid intersection, and carrier oil entered from the top and bottom. A nascent bolus of aqueous liquid was imaged in the intersection just before it broke into separate 4 pL droplets as the fluid tension began to exceed the surface tension of the aqueous liquid. A steady train of droplets exiting the intersection to the right was collected off-chip as a stable emulsion for thermal cycling. FIG. 3b illustrates droplet spacing for readout. The flow was prepared as in 3a, except that the emulsion from (a) was injected into the intersection from the left after thermal cycling instead of the continuous phase. The emulsion appeared close-packed before the intersection in this image because oil was drained from the emulsion during off-chip processing. The droplets were separated by oil introduced at the intersection, and the fluorescence of each droplet was measured at the positions marked by the arrows. Figure 3c shows a cartoon of the droplet readout by fluorescence. The relatively rare PCR(+) droplets (light grey) flow towards the detector together with the majority of PCR(-) droplets (dark grey). The droplets are sequentially interrogated by laser-induced fluorescence while passing through the detection region.

[0076] For serial dilutions, the average number of target DNA molecules per droplet - henceforth referred to as "occupancy" - should increase directly proportional to the DNA concentration. This occupancy was calculated from Poisson statistics using the following equation, well known to those skilled in the art:

[0077]

number

[0078] The droplets were analyzed by fluorescence as they flowed through the readout chip to count the number of PCR(+) and PCR(-) droplets (see Figure 3c). A burst of fluorescence was observed as each droplet passed the detection zone (marked by an arrow in Figure 3b). To account for small differences in fluorescence intensity between runs that may occur due to different chip configurations, etc., each data set was scaled so that the average fluorescence intensity of the blank droplets was 0.1 V. Figure 4a shows the very short duration of a typical trace of a fluorescence burst from an individual droplet for the sample with the highest DNA concentration in the series. PCR(+) and PCR(-) droplets were easily distinguished by their fluorescence intensity. Two large fluorescence bursts peaking at ~0.8 V occurred from the PCR(+) droplets, whereas a smaller burst due to incomplete fluorescence quenching peaked at ~0.1 V in the PCR(-) droplets. A histogram of peak intensities from the full data set revealed two clear populations centered at 0.10 and 0.78 V (Figure 4b). This evidences that the trend evident in the short trace in Figure 4a was stable over a much longer period. Integration of the two populations in Figure 4b yielded a total of 197,507 PCR(+) and 1,240,126 PCR(-) droplets. Thus, the occupancy for this sample was 0.15 by Eqn. 1, which corresponded to an expected occupancy of 0.18 based on a measured DNA concentration of 110 ng / μL. The occupancy for each sample in the serial dilutions was measured and calculated using the dilution equation:

[0079]

number

[0080] Multiplexing primers in digital PCR reactions Droplet-based digital PCR techniques such as those described in Link et al. (US Patent Application Nos. 2008 / 0014589, 2008 / 0003142, and 2010 / 0137163), Anderson et al. (US Patent No. 7,041,481 and reissued as Reissue No. 41,780) and Raindance Technologies Inc. European Patent Application Publication No. 2047910 (the contents of each of which are incorporated herein by reference in their entirety) use a single primer pair per library droplet. This library droplet is merged with a template droplet containing all PCR reagents, including genomic DNA, except for primers. After merging the template droplet with the primer library droplet, the new droplet now contains all the reagents necessary to perform PCR. The droplet is then thermal cycled to produce amplicons. In one embodiment, the template DNA is diluted with the template mix so that there is less than one haploid per droplet on average.

[0081] Having only one haploid genome (i.e., one allele) per droplet gives droplet PCR an advantage over standard singleplex or multiplex PCR in tubes or microwells. For example, in traditional PCR, both alleles are present in the reaction mix, so if there is a difference in PCR efficiency between alleles, the allele with the highest efficiency will be over-represented. In addition, despite careful primer design, there may be mismatches in the sequences to which the PCR primers hybridize. Mismatches in primer hybridization sequences can cause primers to have lower hybridization efficiency for the allele with this mismatch compared to the allele with the wild-type sequence. This can also cause one allele to be preferentially amplified over the other allele when both alleles are present in the same reaction mix.

[0082] Droplet-based PCR avoids these problems because there is only one template molecule per droplet, and therefore one allele per droplet, so even if there is a primer mismatch that reduces PCR efficiency for one allele, there is no competition between the alleles because they are separated and therefore uniformly amplified.

[0083] It is known that traditional multiplexing of standard PCR primers in tubes or wells is difficult to optimize. Multiple PCR amplicons produced in the same reaction can result in competition between amplicons with different efficiencies due to sequence or length differences. This can result in yield variation between competing amplicons, resulting in uneven amplicon yields. However, droplet-based digital PCR utilizes only one template molecule per droplet, so even if there are multiple PCR primer pairs in the droplet, only one primer pair is active. Since only one amplicon is produced per droplet, there is no competition between amplicons, resulting in more uniform amplicon yields between different amplicons.

[0084] A certain amount of DNA is required to generate either a certain amount of DNA to achieve sufficient sequencing coverage per base and / or a certain number of PCR positive droplets. Because only a small percentage of droplets, roughly one-third in standard procedures, are PCR positive, it takes more DNA to achieve an equivalent PCR yield per template DNA molecule. By adding more genomic DNA, the number of PCR positive droplets and therefore the amplicon yield can be increased. For example, increasing the amount of genomic DNA by two-fold while keeping the number of droplets constant will double the amplicon yield. However, there is a limit to the amount of genomic DNA that can be added before there is a significant chance that a gene in the same droplet will have both alleles, and as a result, the advantage of droplet PCR for overcoming allele-specific PCR and resulting in allele dropout is eliminated.

[0085] One way to allow for more genomic DNA input is by generating more droplets and keeping the ratio of haploid molecules per droplet constant. For example, doubling the amount of DNA and doubling the amount of droplets increases amplicon yield by 2X while maintaining the same haploid genome ratio per droplet. However, doubling the number of droplets is problematic, and increasing the amount of DNA is difficult for users with limited amounts of DNA.

[0086] Multiplexing PCR primers within droplets allows for the simultaneous increase of the number of PCR droplets while keeping the input DNA amount the same or less and resulting in comparable or greater amplicon yields. This results in an overall increase in the amount of PCR positive droplets and amplicon yield without consuming more DNA.

[0087] As an example, if there is an average of one haploid genome per four droplets or an average of one-quarter haploid genome per droplet and one PCR primer per droplet, the chance of there being a suitable template for the PCR primer in the droplet is one in four. However, if there are two PCR primers per droplet, the chance of there being a suitable template in the droplet is doubled. This results in one droplet out of two being PCR positive, which doubles the amplicon yield without doubling the input DNA. If the number of droplets containing 2x multiplexing primers is doubled and the DNA is kept constant, the number of PCR positive droplets falls back to one-quarter, but the total number of PCR droplets remains the same because the number of droplets has doubled. If the multiplexing level in each droplet is increased to 4x and the input DNA is the same, the chance of there being a suitable template molecule in each droplet is doubled. This results in a two-fold increase in the number of PCR positive droplets, which doubles the amount of amplicon yield without increasing the amount of input DNA. Thus, by increasing the multiplexing of PCR primers within each droplet, and by increasing the number of droplets overall, amplicon yields can be increased four-fold without increasing the amount of input DNA.

[0088] Alternatively, if the amplicon yield is already sufficient, the amount of input genomic DNA can be reduced without sacrificing amplicon yield by increasing the multiplexing level for the PCR primers in each droplet. For example, going from 1x to 2x the multiplexing level of PCR primers can reduce the amount of input genomic DNA by 2x while still maintaining the same total amplicon yield.

[0089] Even if the number of PCR primer pairs per droplet is greater than one, there is still only one template molecule per droplet, and therefore only one primer pair per droplet will be utilized at a time, meaning that the advantages of droplet PCR in eliminating bias from either allele-specific PCR or competition between different amplicons are maintained.

[0090] An exemplary demonstration experiment of droplet-based amplification and detection of multiple target sequences in a single droplet is presented here. Multiple copies of five sets of primers (primers for TERT, RNaseP, E1a, SMN1 and SMN2) were encapsulated in a single droplet at various concentrations together with template DNA and PCR master mix. Probes that specifically bind to TERT, RNaseP, E1a, SMN1 or SMN2 were also encapsulated in the droplets containing those primers. The probes for TERT, RNaseP and E1a were labeled with VIC dye, and the probes for SMN1 and SMN2 were labeled with FAM dye. The sequences of TERT, RNaseP, E1a, SMN1 and SMN2 were amplified by PCR. PCR was performed with standard thermal cycling settings. For example: 95 °C for 10 min 31 cycles 92°C for 15 seconds 60°C for 60 seconds.

[0091] At the end of PCR, the fluorescence emission from each droplet was determined and plotted in a scatter plot based on its wavelength and intensity. Six clusters were shown, each representing droplets with a corresponding fluorescence wavelength. The TERT, RNaseP and E1a clusters showed VIC dye fluorescence at three distinct intensities, while the SMN1 and SMN2 clusters showed FAM dye fluorescence at two distinct intensities (Figure 5). The number of droplets each carrying one or more sequences selected from TERT, RNaseP, E1a, SMN1 and SMN2 can be determined from the scatter plot.

[0092] In another demonstration experiment of droplet-based amplification and detection of multiple target sequences in a single droplet, five sets of primers (primers for TERT, RNaseP, E1a, 815A and 815G) were encapsulated in a single droplet at various concentrations together with template DNA, PCR master mix and probes. Five different probes, TERT, RNaseP, E1a, 815A and 815G, were also encapsulated in the droplets containing those primers. The probes for TERT and 815A were labeled with VIC dye, and the probe for 815G was labeled with FAM dye. For each of RNaseP and E1a, two probes were encapsulated, one labeled with VIC dye and the other labeled with FAM dye.

[0093] Droplets containing both primers and probes were fused with droplets containing templates. PCR reactions were performed on the fused droplets to amplify the sequences TERT, RNaseP, E1a, 815A and 815G. PCR was performed with standard thermal cycling settings.

[0094] At the end of PCR, the fluorescence emission from each fused droplet was determined and plotted in a scatter plot based on its wavelength and intensity. Six clusters were shown, each representing a droplet with a corresponding fluorescence wavelength and intensity. The TERT and 815A cluster showed VIC dye fluorescence at two distinct intensities; the 815G cluster showed FAM dye fluorescence; and the RNaseP and E1a cluster showed both FAM and VIC dye fluorescence at distinct intensities (Figure 6). The number of droplets each having one or more sequences selected from TERT, RNaseP, E1a, 815A and 815G can be determined from the scatter plot. The copy number of RNaseP, E1a, 815A and 815G in the template was determined by the ratio between the number of droplets having RNaseP, E1a, 815A and / or 815G sequences and the number of droplets having TERT sequences (Figure 6).

[0095] In another exemplary demonstration of multiplexing primer pairs in droplet-based digital PCR reactions, two droplet libraries were generated: droplet library A, in which each droplet contained only one primer pair; and droplet library B, in which primer pairs were multiplexed at a 5x level within each droplet. HapMap sample NA18858 was processed in duplicate with droplet libraries A and B using standard procedures. 2 μg of sample DNA was used for droplet library A, and 1 μg of sample DNA was used for 5x multiplex droplet library B. After PCR amplification, both droplet libraries were disassembled and purified with Qiagen MinElute columns, and then run on an Agilent Bioanalyzer. Samples were sequenced by Illumina with 50 nucleotide reads using Illumina GAII, and their sequencing results were analyzed using standard sequencing metrics. Results from 5x multiplex droplet library B were compared to singleplex droplet library A using standard metrics shown in the table below.

[0096] Results from 5x multiplexed droplet library B were equivalent or better than those from droplet library A. Primer multiplexing produces sequencing results that are identical in base coverage, specificity and uniformity to those achieved by singleplex, with the added benefit of reduced input DNA.

[0097] [ka] Total Reads: The total number of sequencing reads found in a given sample data set. Mapped reads (%): Percentage of total reads that are mapped to the human genome. Specificity: The percentage of mapped reads that contain the target, including all amplicon sequences but not including primer sequences. Average base coverage: The average base coverage in a target, which includes all amplicon sequences but not primer sequences. C1: % of targets with at least 1x base coverage. Note: Non-unique sequencing reads are randomly mapped. C20: % of targets with at least 20x base coverage. C100: % of targets with at least 100x base coverage. Base coverage (0.2x average): % of targets with at least 20% of the average base coverage.

[0098] Single-color gene copy number assay Traditional digital PCR methods involve the use of single-labeled probes specific for individual targets. Figure 7 is a schematic diagram illustrating one-color detection of target sequences using droplet-based digital PCR. As shown in panel A of Figure 7, template DNA is amplified with a forward primer (F1) and a reverse primer (R1). A probe (P1) labeled with a fluorophore of color 1 binds to the target gene sequence (target 1). Microdroplets are created with a diluted solution of template DNA under limiting or end-point dilution conditions. Droplets containing the target sequence emit fluorescence and are detected by a laser (panels B and C). The number of microcapsules containing or not containing the target sequence is shown in a histogram (D) and quantified (E).

[0099] Figure 8 is a schematic diagram illustrating the dual-color detection of two gene sequences in a microfluidic device. As shown in panel A of Figure 8, template DNA is amplified with two sets of primers: forward primer (F1) and reverse primer (R1), and forward primer (F2) and reverse primer (R2). A probe labeled with a fluorophore of color 1 (P1) binds to target 1, and a probe labeled with a fluorophore of color 2 (P2) binds to target 2 (panels B and C). Microdroplets are made with a diluted solution of template DNA under limit or end-point dilution conditions. Droplets containing target sequence 1 or 2 emit fluorescence of color 1 or 2, respectively, and are optically detected by a laser (panels B and C). The number of microcapsules containing target 1 or 2 is shown in panel (D) by a histogram.

[0100] The method of the present invention involves accurate quantification of multiple different DNA targets by dPCR using probes with the same fluorophore. Figure 9 is a schematic diagram illustrating dual-color detection of three gene sequences in a microfluidic device. As shown in panel A of Figure 9, template DNA is amplified with three sets of primers: forward primers (F1, F2 and F3) and reverse primers (R1, R2 and R3). Probes (P1, P2 and P3) are labeled with fluorophores (color 1, color 2 and color 1) and bind to target gene sequences (target 1, target 2 and target 3) (panels B and C). Microdroplets are made with a diluted solution of template DNA under limit or end-point dilution conditions. Microdroplets containing target sequence 1 or 3 emit fluorescence of color 1 at two different intensities; and microdroplets containing target sequence 2 emit fluorescence of color 2. The number of microdroplets containing targets 1, 2 or 3 is shown by a histogram in panel D.

[0101] Previous results from droplet digital PCR (dPCR) demonstrate that multiple independent PCR reactions can be run using the same fluorophore and quantified separately. Notably, the SMN2 assay gives rise to an unexpected droplet population with a significantly higher signal in the FAM detection channel.

[0102] The results are illustrated in FIG. 10. The left dot plot in FIG. 10 illustrates the effect of having an SMN1 blocker present in the reaction. The four clusters illustrated in the left dot plot are as follows: the upper left cluster contains microdroplets containing the reference sequence (SMARCC1); the lower left cluster contains microdroplets that do not contain any sequence; the lower center cluster contains microdroplets that contain the sequence of SMN1; and the lower right cluster contains microdroplets that contain the sequence of SMN2. The right dot plot in FIG. 10 illustrates four clusters where an SMN1 blocker was not present in the reaction; the upper left cluster contains microdroplets containing the reference sequence (SMARCC1); the lower left cluster contains microdroplets that do not contain any sequence; the lower center cluster contains microdroplets that contain the sequence of SMN1; and the lower right cluster contains microdroplets that contain the sequence of SMN2. The shift in the lower center cluster in the right panel compared to the left panel confirms that the fluorescence intensity provides a very sensitive measure of the presence of the sequence.

[0103] Without intending to be bound by any theory, the simplest explanation is that the cluster arises from weak association of the SMN2 probe to the SMN1 gene, despite the presence of a blocker for that gene (a non-fluorescent complementary probe to the SMN1 gene).

[0104] One conclusive confirmation of SMN1 as the source of the unexpected clusters was the observed dependence of the intensity of this signature on the presence of SMN1 blocker. A clear shift to higher FAM fluorescence intensity was observed in the absence of blocker (Figure 10). Another conclusive confirmation was the ratio of SMN1 (estimated) population size to reference size (two copies each) of 0.96 (S_131 sample), in perfect agreement with expectation. Another sample with the same number of SMN1 copies, S_122, yielded a ratio of 0.88 in one run and 0.93 in another, also consistent with the proposed explanation for the unexpected clusters.

[0105] Without intending to be bound by any theory, these observations indicate that binding of the SMN2 probe to SMN1 DNA results in a high fluorescent signal. A simple kinetic model to explain this phenomenon assumes that hybridization of the SMN2 probe to SMN1 DNA reaches equilibrium at a faster rate than the polymerase can occlude the complementary strand. Thus, the amount of probe fluorophore released during each thermal cycle is proportional to (or even equal to) the number of probes bound. Therefore, the lower the binding affinity, the lower the number of probe fluorophores released. Due to the mismatch(es) between the SMN2 probe and SMN1 sequences, one would certainly expect the affinity of the probe to be lower for SMN1 than for SMN2. This model also explains the signal dependence on the sMN1 blocker: the blocker competitively inhibits SMN2 probe hydrolysis by the polymerase exonuclease activity.

[0106] However, there are also cases where probe hybridization does not reach equilibrium before exonuclease activity. In this case, the association rate would play a more dominant role. A similar theory applies. The binding rate to the matching site is likely to be faster than to the mismatch site, and the blocker would act to slow down the probe binding to the mismatch site. Although the binding of the SMN2 probe to SMN1 DNA may be detectable by conventional bulk qPCR, especially in the absence of SMN2, highly quantitative results such as those shown here are highly unlikely. Indeed, there are no reports of qPCR or any other technique quantifying two different DNA sequence motifs with the same color fluorophore. Isolation of individual reactions by single molecule amplification in droplets eliminates any confusion regarding heterogeneous contributions to the signal.

[0107] The advantages of quantifying DNA with multiple probes of the same color fluorophore extend beyond the example of two highly homologous sequences shown here: rather, any number of sequences of any degree of similarity or dissimilarity can be quantified as long as the different probes have significantly different binding occupancies for their respective DNA binding sites.

[0108] Another advantage of the dPCR approach for multiplexed reactions is that different reactions do not compete with each other for reagents as in bulk qPCR assays. However, the possibility for unexpected cross-reactivity remains. Multiplexes assays may require more dilute samples. For example, at 10% occupancy, a duplex reaction will have 1% occupancy twice as much as it does. Therefore, one of the 10 PCR+ droplets will be doubled, resulting in a final intensity at least as high as the brightener of the two probes and possibly higher. For a simple duplex system, the contribution from each probe could be recovered. In this example, the total number of PCR+ droplets for probe 1 would be (probe 1) + (probe 1 + probe 2). Higher multiplexes require greater dilution. For example, for a 4-plex at 1% occupancy, the probability of one probe overlapping with any of the other three is ~3%, an error that is too high for some applications. The need for large dilutions favors large numbers of dPCR reactions.

[0109] In another embodiment of the present invention, a single fluorophore (FAM) was used in gene copy number assays for both reference and target DNA. A model system with various concentrations of plasmid DNA was used to represent the change in target gene copy number relative to the reference gene, corresponding to 0-16 copies of the target gene per cell. BCKDHA and SMN2 plasmid DNA served as reference and target with 1x and 0.5x primers and probes, respectively. With a starting ratio of 8:1 SMN2 to BCKDHA, samples were serially diluted 2x with a solution of BACDHA of the same concentration, varying only the amount of SMN2. The resulting samples were emulsified and subjected to thermal cycling and incubated for 10 min at 4°C for 1 h. 5 More than 10 droplets were analyzed for each sample as described in the previous section. This process was repeated in triplicate.

[0110] The method of the present invention also includes analytical techniques for the identification of a unique fluorescent signature for each probe. In this example of the present invention, histograms of droplet fluorescence intensity for three different template DNA samples are shown in Figure 11a: no template control (dotted line), BCKDHA only (solid line), and 1:1 BCKDHA vs. SMN2 (dash-dotted line). For clarity, both histograms are shown overlapped to highlight similarities for certain peaks, and juxtaposed to each other to reveal all of the features. In the case of 1:1 BCKDHA vs. SMN2, three populations are easily recognized: the most prominent feature appears at 0.08 V, as well as two minor peaks at 0.27 and 0.71 V. Both minor peaks disappeared in the no template control, but the larger one remained, so the most prominent feature at 0.08 V was assigned to the PCR(-) droplet. The peak at 0.71 V was assigned to BCKDHA as it was the only feature occurring with the addition of BCKDHA alone, and then a peak at 0.27 V appeared with the addition of SMN2, completing the assignment. A very small peak not visible on scale in FIG. 11a appeared at .about.0.9 V, which corresponds to droplets occupied by both genes. As an alternative method of the invention, once the distinct peaks were identified, the droplets within each peak were counted according to their possible state (PCR(+) or PCR(-) for BCKDHA or SMN2 or both), and the gene copy number was then determined from the ratio of occupancy. The gene copy number for each sample at the serial dilutions was plotted against the expected value in FIG. 11b (observed ratio of SMN2 to BCKDHA versus expected ratio of SMN2 to BSKDHA) and was fitted very well to a straight line (y=1.01x) over a sufficient range (R 2 = 0.9997, slope 1.01), confirming accurate and precise measurement of the equivalent of 0 to 16 SMN2 copies per cell.

[0111] Detection of alternatively spliced ​​transcripts The same principle was used to detect and count alternatively spliced ​​transcripts. TaqMan assays can be designed that are specific for each exon in the RNA transcript. After the RNA is made into cDNA, it can be packaged in droplets with no more than one copy per droplet. The droplets will also contain multiplexed TaqMan assays for each exon. Each TaqMan assay will contain a different probe, but all the probes will have the same fluorescent dye. The droplets will be thermal cycled to generate a signal for each TaqMan assay. If there are multiple splice variants in the sample, they will contain different numbers of exons depending on the splicing events. The fluorescence intensity of each droplet will be different depending on the number of exons present. By counting the number of droplets with different intensities, it will be possible to identify the presence and abundance of different splice variants in the sample.

[0112] Copy number variants in heterogeneous samples It will be possible to determine whether a heterogeneous sample contains components with different copy levels. If the copy number variants to be assayed are spaced closely enough along a chromosome, DNA from the sample can be fragmented and packaged into droplets at a level of 1 haploid genome equivalent or less per droplet. The droplets will also contain TaqMan assays specific for that copy number variant. The intensity of the signal in each droplet will depend on the number of copy number variants present for that sample. Counting the number of droplets of different intensity will indicate how many cells in a particular sample have what level of copy number variant.

[0113] Tuning TaqMan® Probe Fluorescence Intensity Identification of probes by fluorescence intensity often requires adjustment of the brightness of the probes, especially for higher-plex assays with dense probe patterns. In the previous section, the probes for gene copy number assays produced very well resolved peaks (Figure 11a). Clearly, there is room to accommodate one or many additional probes in copy number assays to resolve the measurements, but a method is required to adjust the fluorescence intensity of the new probes to avoid interference with existing assays. One method of the present invention involves varying both the probe and primer concentrations as a very simple technique to optimize the relative intensities in higher-plex reactions.

[0114] FIG. 12 is a schematic diagram of tuning the intensity of detectable labels to specific targets in a microfluidic device. As shown in panel A of FIG. 12, template DNA is amplified with two sets of primers: forward primers (F1 and F2) and reverse primers (R1 and R2). Probes (P1 and P2) are labeled with a fluorophore of color 1 and bind to target 1 and target 2, respectively. The fluorescence from target 2 is less intense than that from target 1 due to a single base mismatch between P2 and target 2. As shown in panel B, template DNA is amplified with two sets of primers: forward primers (F1 and F2) and reverse primers (R1 and R2) (panel B). The fluorescence from target 2 is less intense than that from target 1 due to the presence of a competing probe 2 that is not labeled with a fluorophore. As shown in panel C, template DNA is amplified with two sets of primers: forward primers (F1 and F2) and reverse primers (R1 and R2). Probes (P1 and P2) are labeled with a fluorophore of color 1 and bind to target 1 and target 2, respectively. The fluorescence from target 2 is less intense than that from target 1 due to the presence of competing probe 2, which is labeled with a different fluorophore.

[0115] Figure 13 shows the probe fluorescence intensity across a dilution series of probes and primers for a different reference gene, ribonuclease P (RNaseP), against a constant amount of genomic DNA from Coriell cell line NA3814 at an occupancy of 0.02 target DNA molecules per droplet. The probe fluorescence intensity varied linearly with the probe concentration over a narrow concentration range spanning 0.15 to 0.4 μM - roughly centered around a typical probe concentration of 0.2 μM - after correction for dilution errors and other run-to-run differences, e.g., optical realignment using the PCR(-) droplet as a reference (R 2 = 0.995). In summary, probe intensity can be varied by dilution over a small but reasonable range for tuning of multiplexed assays without affecting the amplification itself.

[0116] The above example for adjusting probe fluorescence intensity includes changing both probe and primer concentration at the same rate, but the present invention is not limited to this method for changing probe intensity. Other methods known to those skilled in the art for changing probe intensity are also contemplated. Such methods include changing only probe concentration; changing only primer concentration; changing only forward primer concentration; changing only reverse primer concentration; changing probe, forward and reverse primer concentration in any way; changing thermal cycling program; changing PCR master mix; incorporating any part of the probe that lacks fluorophore into the assay; or incorporating any hybridization-based competitive inhibitor of probe binding, such as blocking oligomeric nucleotides, peptide nucleic acid and locked nucleic acid, into the assay. The present invention incorporates these methods of adjusting probe fluorescence intensity, used alone or in any combination, or any other method for adjusting probe fluorescence intensity.

[0117] Higher plex reactions One method of the present invention involves performing higher plex assays using a single probe color (i.e., fluorophore). As explained above, various means can be used to adjust the probe fluorescence intensity so that each intensity level uniquely identifies a DNA target. For example, targets T1, T2, T3, and T4 can be uniquely identified by intensity levels I1, I2, I3, and I4. Without intending to be bound by theory, the maximum number of intensity levels possible for unique identification of a target is related to the resolution of the different intensity levels - i.e., the width of the intensity for each specific probe compared to the gap between the average intensities of the probes - as well as the intensity of the empty droplets, which tends to increase as the number of probes increases. The number of intensity levels can be 0, or 1, or 2, or 3, or 4, or 10 or less, or 20 or less, or 50 or less, or 100 or less. The number of intensity levels can be more than 100. In the examples shown below, as many as 3 intensity levels are demonstrated.

[0118] Another method of the present invention involves performing a higher plex assay using multiple different probe colors (i.e., fluorophores). As with the single-color multiplexed assay above, for each color probe, multiple targets can be identified based on intensity. In addition, multiple colors that are spectrally separable can be used simultaneously. For example, a single droplet may contain four different probes for four different targets. Two probes may be of color A (e.g., A1 and A2) with different intensities, and the other two probes may be of color B (e.g., B1 and B2) with different intensities. The corresponding targets are T1, T2, T3, and T4 for A1, A2, B1, and B2, respectively. If a droplet showed an increase in fluorescence of color A, the droplet contained either target T1 or T2. Then, based on the fluorescence intensity of color A, the target could be identified as T1 or the target could be identified as T2. However, if a droplet showed an increase in fluorescence of color B, the droplet contained either target T3 or T4. Then, based on the fluorescence intensity of color B, the target could be identified as T3 or the target could be identified as T4. Without intending to be bound by theory, the maximum number of different colors possible is limited by the spectral overlap between the fluorescence emissions of the different fluorophores. The maximum number of colors can be 1, or 2, or 3, or 4, or 10 or less, or 20 or less. The maximum number of colors can be higher than 20. In subsequent demonstration experiments, the maximum number of colors is 2.

[0119] Another method of the present invention involves performing a higher plex assay using multiple different probe colors (i.e., fluorophores), but unlike the above strategy of identifying each target with a single type of probe with a unique color and intensity, this method instead allows for the identification of a single target by multiple probes that constitute a unique signature of both color and intensity. For example, a single droplet may contain four different probes to measure three different targets (e.g., T1, T2, and T3). Two probes may be of color A (e.g., A1 and A2) and two probes may be of color B (e.g., B1 and B2). T1 is measured by probe A1 and T2 is measured by probe B1, but T3 is measured by both probes A2 and B2. Thus, when a droplet contains only T1, an increase in fluorescence of color A appears. When a droplet contains only T2, an increase in fluorescence of color B appears. However, when a droplet contains T3, an increase in fluorescence of both colors A and B appears.

[0120] In general, without wishing to be bound by theory, the above three methods for higher plex dPCR are easiest to implement under conditions of end-point dilution, i.e., when the probability that multiple different target molecules will occupy the same sample together is very low compared to the probability that any single target will occupy a droplet. Multiple occupancy creates the complexity of competing simultaneous assays in the same reaction droplet, and also the complexity of the allocation of the resulting fluorescence intensity, including the combination of fluorescence from two different reaction products that may or may not be equal to the sum of the two fluorescence intensities of the individual reaction products. However, the method of the present invention can accommodate these complications resulting from multiple occupancy.

[0121] The method of the present invention for higher plex reactions also includes a method of pairing primers and probes. In the simplest case, for example, when targets are from different cells; or when targets are from different chromosomes in a single cell type; or when targets are far apart from each other in a single chromosome, so that they become physically separated during DNA fragmentation; or when targets are very close to each other in a chromosome, but nevertheless become separated by targeted cleavage of DNA, for example by restriction enzyme digestion; or for any other reason, targets are unlikely to be on the same DNA fragment. In such cases, each probe can be paired with a single primer (forward and reverse) set. However, in other cases, for example, when targets are within the same codon, or for any other reason, target regions may often be on the same DNA fragment. In such cases, a single primer set will be useful for multiple probes (see Pekin et al., for example).

[0122] Higher multiplex reactions can be performed to distinguish between the haplotypes of two SNPs. For example, assume that at position 1 there can be genotypes A or A' and at position 2 there can be genotypes B or B'. In a diploid genome, four unique haplotypes are possible (A, B; A, B'; A', B; and A', B'). For example, if A' and B' represent drug resistance mutations for an infection, A'B and AB' are often less severe and treated differently than A'B', which represents significant drug resistance that must be treated with great care. Digital PCR with intensity discrimination is ideally suited for the identification of the low frequency A'B' in the background of a mixture of the other three haplotypes. Haplotyping information is also important for the construction of haplotypes in HLA. One way this example can be constructed is by an assay design where color 1 is used for A, with color 1 being of high or low intensity indicating alleles A or A', respectively, and color 2 is used for B, with color 2 being of high or low intensity indicating alleles B or B', respectively. The population of [color 1, color 2] corresponding to [low, low] is a measure of the AB alleles, and the [high, low] alleles A'B and [A'B'] alleles will be easily distinguishable from [high, high] even in a background that is primarily a mixture of A'B and AB'. See FIG. 22. In some cases it will be advantageous to start by encapsulating a long single molecule of nucleic acid containing both the A and B SNP positions in a droplet, while in other cases it will be desirable to start by encapsulating a single cell, bacterium or other organism in a droplet and then releasing the nucleic acid from the organism. In still other embodiments, multiplex intensity detection of multiple simultaneous targets can be used as a surrogate marker for multiple types of binding interactions or labels of the target material. This technique is also not limited to single molecule detection and can be used for haplotype detection in single cells (e.g., bacteria, somatic cells, etc.) For single cell analysis, a sorting step may be applied prior to haplotyping.

[0123] 5-plex assay for spinal muscular atrophy One aspect of the present invention is summarized in the implementation of a case-demonstration study of the quantification of several genetic markers for spinal muscular atrophy (SMA). SMA was selected for one of the case-demonstration studies because of both its significant clinical utility and its complex genetic nature. It is the second most frequent fatal neurodegenerative disease, affecting ∼1 in 10,000 births. SMA is most often caused by homozygous absence of exon 7 in the survival motor neuron 1 gene (SMN1, reviewed by Wirth et al.), but the severity of the condition is modulated by the number of gene copies of SMN2, with prognosis ranging from lethal to asymptomatic at copy numbers of 1-5 (reviewed by Elsheikh et al.). Accurate quantification of SMN2 copy number is therefore important for clinical prognosis and genetic counseling. Apart from large deletions of SMN1, a number of point mutations or short deletions / duplications within the same gene are also responsible for ∼4% of SMA cases. In a significant step towards a comprehensive SMA assay, the multiplexed dPCR assay demonstrated here includes both a copy number assay (for SMN1 and 2) and an assay for a common SNP (c.815A>G).

[0124] One embodiment of the present invention is a 5-plex assay for SMA diagnosis. This 5-plex assay quantifies common genetic variants that affect SMA and includes two copy number assays for SMN1 and SMN2 genes with BCKDHA as a reference and an SNP assay for c.815A>G mutation. Two different color fluorophores, FAM and VIC, were used to uniquely quantify each of the assays. The probes for SMN1 and SMN2 contained only FAM, and the probe for c.815A contained only VIC. However, a mixture of VIC and FAM labeled probes was used for BCKDHA and c.815G. The use of VIC and FAM fluorophores in this example does not limit the present invention, rather, the 5-plex assay can be used with any spectrally separable fluorophore compatible with TaqMan assays or with any other fluorogenic hybridization-based probe chemistry. To validate the assay, model chromosomes were synthesized containing a single target region for each of the different primer / probe pairs. EcoRV restriction sites flanked each target, allowing the fragments to be separated.

[0125] As another method of the present invention, histogram-based data presentation and analysis is incorporated into the present invention for identification and characterization of statistically similar populations of droplets resulting from one probe signature (color and intensity) and for distinguishing one droplet population from another. Figure 14a shows a 2D histogram of droplet fluorescence intensity as a contoured heat map, with warmer colors representing higher occurrence. Standard techniques were used to correct for spectral overlap of FAM and VIC signals. Samples were run at 0.006 occupancy per target. Six populations, 5 for assay and 1 for PCR(-) droplets, are clearly visible. As a method of the present invention, the populations were assigned by selective elimination of assay components. For example, elimination of SMN2 primers and probes eliminated the population at the bottom right of the histogram, but otherwise the distribution remained unchanged. The assignments are labeled in Figure 14a. We found this multiplexing method to be generally applicable, so we readily assayed well-resolved or at least distinguishable populations of each target. Another method of the present invention then adjusted the relative positions of the different populations in the histogram into a regularly spaced rectangular array by tuning the probe concentrations as described in the previous section. Typically, no more than two iterations were required for optimization.

[0126] In another method of the present invention, the different populations were resolved well enough to allow counting of droplets within each population by integration across a rectangular boundary. The boundary is located midway between adjacent peaks. The method of the present invention is not bound to a rectangular boundary and a specific boundary location between peaks. Rather, any closed or open boundary condition is sufficient. The boundary conditions do not have to be "binary" in the sense that weighted integration can also be performed across the boundary to arrive at the droplet count. The peak location of each cluster does not vary more than 2% between runs after normalization to the intensity of empty droplets to account for variations in detection efficiency (data not shown). Thus, once identified, the same boundary could be reused between samples for integration. The method of the present invention is not limited to fixed boundary locations. Dynamic population identification and boundary selection across samples or studies is pre-processed. 20 different patient samples from the Coriell cell repository were analyzed with this assay: 4 affected by SMA, 1 SMA carrier, and 15 negative controls. The assay results are shown in Figure 14b. Gene copy numbers were calculated as before as the ratio of occupancy derived from the number of target droplets versus the reference droplets. As with the copy number measurements in Figure 11, each assay yielded ratios very close to the expected integer values, but when all of the patient data were plotted as actual to expected integer ratios, small systematic deviations from an ideal slope of 1 were observed. The actual slopes were 0.92, 0.92, and 0.99 for SMN1, SMN2, and c.815A, respectively. For clarity, the data in Figure 14b have been scaled according to an ideal slope of 1.

[0127] The observed genotypes for the different patients were consistent with their disease status (unaffected, carrier, or affected). Each patient affected with SMA had zero copies of SMN1 (numbers SMA1-4 in Fig. 14b), the carriers had just one copy, and all the negative controls had two or three copies (numbers 1-15). Three unrelated individuals (numbers 6, 8, and 9) had three copies of SMN1, which occurred at a rate of 20%, similar to previous reports for healthy individuals. The variability in SMN1 copy number is not surprising since it resides within an unstable region of chromosome 5q13. A greater variety of SMN2 copy numbers was observed. One to two copies were most common in the control group, while one individual had zero copies (a distribution consistent with the expected value for normal individuals). SMA carriers and affected patients had high copy numbers of SMN2 on average: carriers had 5; two affected had 3 copies, and the others had 2 copies. All affected patients were diagnosed with SMA type I, the most severe form, based on clinical observations from the Coriell repository. The strong genotype / phenotype correlation between SMN2 copy number and disease severity suggests that the two individuals with 3 copies of SMN2 may have an improved type II prognosis, especially with respect to patient SMA1, who had lived more than 3 years at the time of sampling, well beyond the typical maximum life expectancy for SMA type I of 2 years. However, one is still hesitant to predict disease outcome based on SMN2 copies alone, as other less well-characterized or unknown modifier genes may affect prognosis, and because not all SMN2 copies may be full genes. Moreover, some type I patients are beginning to survive longer in newer clinical settings. Thus, with the little clinical information available to us about the patients, our SMN2 assay results were consistent with broad expectations about disease severity.

[0128] The SNP assay showed that all patients carried the normal c.815A genotype, and no cases of c.815G were observed. The mutation is relatively rare and therefore was not expected to appear in the small patient panel. However, the presence of an apparent extra gene fragment in two unrelated individuals not covered by the SNP assay was intriguing. The c.815A>G assay does not distinguish between SMN1 and SMN2 due to their high sequence similarity, and therefore the total copies of c8.15A and G would be equal to the sum of the copies of SMN1 and SMN2. This was true for all patients except for healthy patients, numbers 1 and 2, who both had one extra copy of c.815A. Since c.815 resides on exon 6 and the SNP that distinguishes the SMN1 and SMN2 genes resides on exon 7, the extra gene could be a fragment of SMN1 lacking exon 7. This seems plausible because the deletion of exon 7 is a common mutation responsible for 95% of SMA cases, with 1 / 40 to 1 / 60 adults carrying it, and therefore these patients would have been typical carriers of SMA had they not had at least one compensatory gain of a healthy copy of SMN1 on the same chromosome.

[0129] 9-plex assay for spinal muscular atrophy A 9-plex assay for certain SMA-related targets was also demonstrated with only two colors (probes containing FAM and VIC fluorophores). Except for the optimized primer and probe concentrations, the assay conditions and experimental procedures were identical to the 5-plex assay described above. Figure 15a shows the various droplet populations before optimization of the probe concentration in a 2D histogram. The identity of the different targets is shown in the figure itself. As a method of the invention, the identification of the different populations was performed by selective exclusion and / or addition of one or more assays as before. Most of the populations were already well resolved, except for the probe for the c.815A genotype, which was in close proximity to the cluster corresponding to the empty droplets. After three iterations of optimization of the probe concentration, all target populations were well resolved from each other and from the empty droplets (Figure 15b). This demonstration highlights three methods of the invention: (1) nine DNA targets were uniquely identified in two-dimensional histograms, far beyond the capabilities of conventional qPCR; (2) target DNA molecules were differentiated based on some combination of both color and intensity resulting from one or more probes for the same target; and (3) the relative positions of target molecules in the histograms were adjusted by varying probe concentrations to optimize color and intensity patterns for increased resolution among various droplet populations.

[0130] As one method of the present invention, different droplet populations are identified by selectively adding or removing assays in the above examples. However, the present invention is not limited to this method. Rather, any method of population allocation known to those skilled in the art is considered. The method of the present invention includes any method that can cause identifiable replacement, appearance or disappearance of one or more populations in the histogram, including changing both probe and primer at the same or different rates; changing only probe concentration; changing only primer concentration; changing thermal cycling conditions; and changing master mix composition. Another method of the present invention utilizes previous knowledge of the position of assays in the histogram to assist in allocation.

[0131] Multiplexing ability The level of multiplexing demonstrated in the SMA example was 9x, significantly exceeding the maximum feasible number for qPCR. Without wishing to be bound by theory, the two main limitations are inter-assay resolution and the increasing fluorescence intensity of empty droplets with higher probe loading. The method of the present invention involves optimizing the color and intensity patterns of different probes for maximum multiplexing while still achieving appropriate specificity for each individual reaction. Although rectangular arrays of droplet populations were demonstrated for 5- and 9-plex reactions, another desirable pattern is a tightly packed hexagonal array. However, the present invention is not bound to any particular array strategy.

[0132] Adding extra colors would increase that possibility even further, but with some diminishing returns as the fluorescence of empty droplets continues to intensify. This capacity can be increased even further with better probes that produce larger differential signals, for example hybrid 5'-nuclease / molecular beacon probes that reduce background by contact quenching but exhibit a bright signal characteristic of free unquenched fluorophores. With such improvements, multiplexing capabilities of over 50x can be envisioned.

[0133] Combined multiplexing with optical labeling Droplet-based microfluidics can also be used to simultaneously measure multiple targets by different methods. According to an alternative method, primers and probes can be individually loaded into droplets along with optical labels to uniquely identify the assays. Typically, the optical labels are a fluorophore or a combination of different fluorophores that are spectrally distinct from the probe fluorophore. A variety of different types of droplets can be mixed into a "library" of droplets, each containing a different assay that is uniquely identified by a different optical label. The template DNA-containing droplets and library droplets are then combined one by one according to the method of the present invention above. After thermal cycling, some droplets containing template DNA exhibit brighter fluorescence at the emission wavelength of the probe. The specific target DNA molecules that give rise to these PCR(+) signals are then identified by the optical probe. In one study, six common mutations in KRAS codon 12 were screened in parallel in a single experiment by fusing droplets containing genomic DNA with one of seven different types of droplets (7-member libraries), each containing TaqMan® probes and optical codes specific for a different KRAS mutation or wild-type KRAS.

[0134] In one method of the invention, optical labeling can be used in conjunction with various multiplexing methods of dPCR already incorporated in the invention. For example, a single optical label can code not only a single assay as in the KRAS example above, but the entire 5-plex SMA assay. In this manner, other optical labels can code various screening assays for newborns. In accordance with other methods of the invention above, a single DNA sample from an infant and all the assays could then be analyzed simultaneously by combining the droplets containing the DNA with the library droplets containing the optically encoded assays one by one.

[0135] As an example of using multiplexing in conjunction with optical labeling, we demonstrated a so-called 3x3x3 combination multiplex reaction and optical labeling (3x3 optical labeling with two fluorophores, each encoding a triplex assay, for a total of 27-plex). Two fluorophores, Alexa633 and CF680 (excited by a 640 nm laser), each with three intensity levels resulting in a total of nine optical labels, were utilized for optical labeling. As in the previous 5- and 9-plex assays for SMA, TaqMan assays were used with FAM and VIC fluorophores (excited by a 488 nm laser). Fluorescence from the FAM and VIC fluorophores was recorded simultaneously with fluorescence from the optical labels, which required modifications to the optical layout of the instrumentation described for the SMA assay (the optical schematic for two-laser excitation and four-color detection is generally shown in Figure 16). Also, in this example, co-flow microfluidics was used (the use of co-flow based microfluidics for this application is one of the methods of the invention described above). In this case, template DNA was introduced to the chip in one flow, and PCR master mix, primers and probes for one triplex assay and a unique composition of fluorophores for optical labeling were simultaneously introduced to the chip in the other flow. The two flow streams converged at a fluidic intersection upstream of the droplet formation module, and each droplet thus formed contained the contents of both flow streams. Methods for performing co-flow microfluidics are well known to those skilled in the art. The droplets were collected and then the procedure was repeated with the next triplex assay and optical label. The procedure was repeated nine times, once for each pair of assay and optical label. All droplets were collected in one PCR tube and subjected to thermal cycling off-chip. The thermocycled droplet mixture was re-injected into the same readout chip used above for the SMA assay, and the fluorescence intensities of the assays from all four fluorophores were recorded.

[0136] Figure 17 shows the cumulative results from all droplets in a 3 x 3 x 3 assay using co-flow microfluidics. This figure shows 2D histograms of droplet fluorescence intensity, from all optical labels on the left, and from the assays on the right. Standard methods were used to correct for spectral overlap in the optical label histograms. These histograms are shown as heat maps, with warmer colors representing larger droplet numbers. Nine different clusters of droplets were clearly visible in the optical label histograms. Each of these nine corresponds to a different optical label: there is a small group of four clusters in the bottom left corner of the histogram, corresponding to the optical label with the lowest fluorescence intensity; as well as five clusters that appear as linear streaks at higher intensity. The droplet clusters were less clear in the histogram for the assays, which was expected since the droplets shown contain all of the triplex assays. Selecting a single type assay by using optical labels, as described below, allowed the individual assays to be clearly defined.

[0137] The methods of the invention include selecting individual populations of droplets that all contain the same optical label, or group of optical labels. In some methods of the invention, fluorescence intensity boundaries are used to identify the populations. In the examples shown, rectangular boundaries are used to identify minimum and maximum fluorescence intensities for each fluorophore. However, the methods of the invention are not constrained to rectangular boundaries. Any boundary, closed or open, can be used. Furthermore, according to the methods of the invention, the selection of droplet populations can be done by any method and is not constrained to threshold-based methods such as boundary selection.

[0138] FIG. 18A shows the droplet fluorescence intensity of the assays (right histogram) when only one optical label was selected (left histogram). The overlapping lines on the histograms of the optical labels confirm the rectangular boundaries used to select only the optical labels with the lowest fluorescence for both fluorophores. Both histograms showed only the selected droplets. After selection, four distinct clusters of droplets appeared in the assay histograms, three for the different assays (in this case assays for SMN1, SMN2 and TERT (TERT is another common reference gene)) and one for the empty droplet. The copy numbers for SMN1 and SMN2 were measured by the same method of the invention as described above for the 5-plex SMA assay, with values ​​of 1.8 and 0.94, close to the expected values ​​of 2 and 1, respectively. The same assays were coded with two other optical labels. Their selection is shown in FIG. 18B and C. Similar results were achieved, with combined measurements of 1.9±0.1 and 0.9±0.1 copies of SMN1 and SMN2, respectively, indicating precise measurements within experimental uncertainty.

[0139] Figures 19A, B and C show the optical label selection for the different assays (TERT, c.5C in the SMN1 gene, and BCKDHA (labeled E1a in this figure)). In each case, four different clusters also appeared, and the same inventive method as above gave accurate measurements of gene copy numbers for c.5C and BCKDHA with reference to TERT of 2.9±0.1 and 2.0±0.2 compared to 3 and 2, respectively. Figures 20A, B and C show the optical label selection for the third assay (TERT, c.88G in the SMN1 gene, and RNaseP (RNaseP is the common reference gene)). An accurate gene copy number of 2.1±0.1 was measured for both c.88G and RNaseP with reference to TERT, compared to an expected value of 2.

[0140] In summary, the experiments demonstrated herein show the use of nine different optical labels that allow independent measurement of three triplex assays in a single experiment. Although in this example some of the optical labels code for overlapping assays (only three different assays despite having nine optical labels), the present invention is not limited to any particular format of assays and optical labels. Embodiments of the present invention include formats in which all of the assays are the same across all of the optical labels; formats in which no assay is the same across all of the optical labels; formats in which some of the assays are the same across all of the optical labels; formats in which some of the assays have a higher plexity across all of the optical labels than others; formats in which all of the assays have the same plexity across all of the optical labels; and formats in which any other combination of assays across all of the optical labels is possible.

[0141] Although in this example two different fluorophores were used to create the optical labels, the invention is not limited to any particular number of fluorophores that comprise the optical labels. Embodiments of the invention include optical labels that are comprised of one fluorophore, or two fluorophores, or three fluorophores, or four fluorophores, or ten or fewer fluorophores, or twenty or fewer fluorophores. Optical labels may also include more than twenty fluorophores.

[0142] Although only triplex assays were used in the examples demonstrated herein, the invention is not limited to the use of triplex assays with optical labels. Embodiments of the invention include the following amounts of plexity when used with optical labels: singleplex, duplex, triplex, 4plex, 10plex or less, 20plex or less, 50plex or less, and 100plex or less. Embodiments of the invention also include plexities greater than 100 when used with optical labels.

[0143] Another method of the present invention is to use a non-coflow system to combine multiplexing and optical labeling. A demonstration experiment using droplet merging was performed with the same 3x3x3 assay as in the previous example in co-flow. First, the assays (probes and primers) to be used with their unique optical labels were encapsulated in droplets along with a PCR master mix. Then, following the method of the invention described above, the library containing a mixture of droplets from all nine optically labeled assays was merged one by one with droplets containing template DNA from the same patient as in the previous example. Another method of the invention performed droplet merging using a lambda-injector type merge module as described in U.S. Provisional Patent Application No. 61 / 441,985, which is incorporated herein by reference. Apart from the differences between co-flow and merging, the assays and experimental procedures were identical to those of the co-flow experiments above. Figure 21 shows 2D histograms of droplet fluorescence intensity for optical labels and assays similar to those in Figures 17-20. As with co-flow, the expected distinct clusters of droplets corresponding to each assay were clearly visible, based on the selection of droplets containing individual optical labels. Furthermore, for each assay, the measured gene copy numbers matched or very closely matched the expected values ​​within experimental uncertainty (see Table 1).

[0144] [Table 1] The methods of the invention include the use of either microfluidics with co-flow or microfluidics with droplet merging, although the invention is not limited in this respect. Any fluidics method capable of generating optically labeled droplets that also contain fluorogenic DNA hybridization probes is contemplated. For example, other embodiments known in the art are mixing the optical label and assay in a microfluidic environment prior to injection into a droplet generation chip; and thoroughly mixing the optical label and assay upstream of the droplet formation module in a dedicated mixing module, such as in a serpentine mixer.

[0145] Data analysis One method of the invention includes histogram-based data presentation and analysis to identify and characterize populations of statistically similar droplets resulting from unique probe signatures (color and intensity) and to distinguish one population of droplets from another. Another method of the invention includes histogram-based data presentation and analysis to identify and select populations of droplets based on unique signatures from optical labels. Although examples of one- and two-dimensional histograms for these methods have been provided, the invention is not limited in this respect. As explained above, it is expected that more colors will be used for both multiplexing and optical labels. Thus, embodiments of the invention include histograms with a dimensionality greater than 2, for example 3, or 4, or 10 or less, or 20 or less. Histograms with a dimensionality greater than 20 are also incorporated into the invention.

[0146] Another method of the invention involves the selection of droplets in a histogram for counting, or for assay selection, such as in the case of optical labeling, or for any other purpose. The method of the invention includes selection by boundaries, closed or open, of any possible shape and dimension. The method of the invention also includes the selection of droplets exhibiting fluorescence, either from a single type of fluorophore or from multiple types of fluorophores, e.g., resulting from multiple probes against a common DNA target.

[0147] Polymerase Error Correction For applications requiring ultra-high sensitivity, such as searching for rare mutations in abundant wild-type DNA, false positive results can result from errors from the DNA polymerase itself. For example, during one of the initial thermal cycles, the polymerase may synthesize a mutant strand of DNA from the wild-type template. This type of error is most likely to occur when the difference between the mutant and wild-type is very small, such as a single nucleotide polymorphism (SNP). In this method of the invention, each droplet contains only a single target nucleic acid, if any. In a preferred embodiment, this is accomplished under end-point dilution conditions. Droplets containing an amplification product that is a wild-type target are detected based on the emission of a fluorophore emitted from a probe that hybridizes to the wild-type target. Droplets containing a mutant target are detected based on the emission of a fluorophore emitted from a probe that hybridizes to the mutant target. Since each droplet starts with only a single nucleic acid molecule, the resulting amplification products within each droplet are either uniform for the target or uniform for the target mutants.

[0148] However, a given droplet will contain a heterogeneous mixture of both target and target variants due to polymerase errors during the PCR reaction. The error rate of PCR varies according to the exact nucleic acid sequence, the thermostable enzyme used, and the in vitro DNA synthesis conditions. For example, the error frequency (mutations per nucleotide per cycle) during PCR catalyzed by the thermostable Thermus aquaticus (Taq) DNA polymerase is -2 × 10 -4 to <1×10 -5The frequency of polymerase-mediated errors, at a frequency of 1 mutation per 10,000 nucleotides per cycle, is an important consideration for any PCR application that begins with small amounts of starting material (e.g., less than 10,000 nucleotides total of target DNA) or focuses on individual DNA molecules in the final PCR population.

[0149] The proportion of DNA molecules that contain a sequence alteration is a function of the error rate per nucleotide per cycle, the number of amplification cycles, and the starting population size. The population of modified DNA molecules arises during PCR from two sources: (1) new errors at each PCR cycle; and (2) amplification of error-containing DNA molecules from the previous cycle. The formula

number

number

number

[0150] Polymerase errors that convert wild-type targets to mutant targets during initial amplification rounds can result in heterogeneous populations of targets and target mutants within the droplets, as well as lead to droplets being inaccurately identified as containing mutants of the target (i.e., false positives). Such false positives have a significant impact on the validity and accuracy of digital PCR results.

[0151] The method of the present invention can detect which droplets contain a heterogeneous population of molecules and can exclude them from analysis. As the droplets containing the amplification products flow through the channel through the detection module, the module can detect the fluorescence emission in each droplet. Droplets that produce only a single signal are classified as droplets containing a homogeneous population of targets. Since the probes that hybridize to the wild-type target are labeled with a different fluorophore than the probes that hybridize to the mutants of the wild-type target, the method of the present invention can classify each droplet as containing either a homogeneous population of target amplicons or a homogeneous population of target mutant amplicons.

[0152] Droplets that give rise to two signals are classified as droplets containing a heterogeneous population of molecules. Since each droplet starts with at most a single target nucleic acid, droplets that contain amplification products that are both a target amplicon and a target variant amplicon indicate that the target variant was produced by a polymerase error during the PCR reaction, most often during an early cycle of the PCR reaction. Such droplets are detected and excluded from the analysis.

[0153] analysis Then, analysis is performed only on the droplets that contain a homogeneous population of molecules. This analysis can be based on counting, i.e., determining the number of droplets that contain only wild-type target and determining the number of droplets that contain only mutants of said target. Such methods are well known in the art. See, for example, Lapidus et al. (U.S. Patent Nos. 5,670,325 and 5,928,870) and Shuber et al. (U.S. Patent Nos. 6,203,993 and 6,214,558), the contents of each of which are incorporated herein by reference in their entirety.

[0154] Generally, the presence of droplets containing only the variant is indicative of a disease such as cancer. In certain embodiments, the variant is an allelic variant, such as an insertion, deletion, substitution, translocation, or single nucleotide polymorphism (SNP).

[0155] Cancer-associated biomarkers are known in the art. Biomarkers associated with the development of breast cancer are shown in Erlander et al. (US 7,504,214), Dai et al. (US 7,514,209 and 7,171,311), Baker et al. (US 7,056,674 and US 7,081,340), Erlander et al. (US 2009 / 0092973). The contents of each of the above patent applications and patents are incorporated herein by reference in their entirety. Biomarkers associated with the development of cervical cancer have been described by Patel (US 7,300,765), Pardee et al. (US 7,153,700), Kim (US 6,905,844), Roberts et al. (US 6,316,208), Schlegel (US 2008 / 0113340), Kwok et al. (US 2008 / 0044828), Fisher et al. (US 2005 / 0260566), Sastry et al. (US 2005 / 0048467), Lai (US 2008 / 0311570) and Van Der Zee et al. (US 2009 / 0023137). Biomarkers associated with the development of vaginal cancer are presented in Giordano (US 5,840,506), Kruk (US 2008 / 0009005), Hellman et al. (Br J Cancer. 100(8):1303-1314, 2009). Biomarkers associated with the development of brain cancers (e.g., glioma, cerebellar, medulloblastoma, astrocytoma, ependymoma, glioblastoma) are presented in D'Andrea (US 2009 / 0081237), Murphy et al. (US 2006 / 0269558), Gibson et al. (US 2006 / 0281089), and Zetter et al. (US 2006 / 0160762). Biomarkers associated with the development of renal cancer have been described by Patel (US 7,300,765), Soyupak et al. (US 7,482,129), Sahin et al. (US 7,527,933), Price et al. (US 7,229,770), Raitano (US 7,507,541), and Becker et al. (US 2007 / 0292869).Biomarkers associated with the development of liver cancer (eg, hepatocellular carcinoma) have been presented by Horne et al. (US 6,974,667), Yuan et al. (US 6,897,018), Hanausek-Walaszek et al. (US 5,310,653), and Liew et al. (US 2005 / 0152908). Biomarkers associated with the development of gastric, gastrointestinal and / or esophageal cancer have been reported by Chang et al. (US 7,507,532), Bae et al. (US 7,368,255), Muramatsu et al. (US 7,090,983), Sahin et al. (US 7,527,933), Chow et al. (US 2008 / 0138806), Waldman et al. (US 2005 / 0100895), Goldenring (US 2008 / 0057514), An et al. (US 2007 / 0259368), Guilford et al. (US 2007 / 0184439), Wirtz et al. (US 2004 / 0018525), Filella et al. (Acta Oncol. 33(7):747-751, 1994), Waldman et al. (US 6,767,704), and Lipkin et al. (Cancer Research, 48:235-245, 1988). Biomarkers associated with the development of ovarian cancer have been reported by Podust et al. (US 7,510,842), Wang (US 7,348,142), O'Brien et al. (US 7,291,462, 6,942,978, 6,316,213, 6,294,344, and 6,268,165), Ganetta (US 7,078,180), Malinowski et al. (US 2009 / 0087849), Beyer et al. (US 2009 / 0081685), Fischer et al. (US 2009 / 0075307), Mansfield et al. (US 2009 / 0004687), Livingston et al. (US 2008 / 0286199), Farias-Eisner et al. (US 2008 / 0038754), Ahmed et al. (US 2007 / 0053896), Giordano (US 5,840,506), and Tchagang et al. (Mol Cancer Ther, 7:27-37, 2008).Biomarkers associated with the development of head and neck and thyroid cancer are set forth in Sidransky et al. (US 7,378,233), Skolnick et al. (US 5,989,815), Budiman et al. (US 2009 / 0075265), Hasina et al. (Cancer Research, 63:555-559, 2003), Kebebew et al. (US 2008 / 0280302), and Ralhan (Mol Cell Proteomics, 7(6):1162-1173, 2008), the contents of each of the foregoing articles, patents, and patent applications being incorporated herein by reference in their entirety. Biomarkers associated with the development of colorectal cancer have been reported by Raitano et al. (US 7,507,541), Reinhard et al. (US 7,501,244), Waldman et al. (US 7,479,376); Schleyer et al. (US 7,198,899); Reed (US 7,163,801), Robbins et al. (US 7,022,472), Mack et al. (US 6,682,890), Tabiti et al. (US 5,888,746), Budiman et al. (US 2009 / 0098542), Karl (US 2009 / 0075311), Arjol et al. (US 2008 / 0286801), Lee et al. (US 2008 / 0206756), Mori et al. (US (US 2008 / 0081333), Wang et al. (US 2008 / 0058432), Belacel et al. (US 2008 / 0050723), Stedronsky et al. (US 2008 / 0020940), An et al. (US 2006 / 0234254), Eveleigh et al. (US 2004 / 0146921), and Yeatman et al. (US 2006 / 0195269).Biomarkers associated with the progression of prostate cancer have been reported by Sidransky (US 7,524,633), Platica (US 7,510,707), Salceda et al. (US 7,432,064 and US 7,364,862), Siegler et al. (US 7,361,474), Wang (US 7,348,142), Ali et al. (US 7,326,529), Price et al. (US 7,229,770), O'Brien et al. (US 7,291,462), Golub et al. (US 6,949,342), Ogden et al. (US 6,841,350), An et al. (US 6,171,796), Bergan et al. (US 2009 / 0124569), Bhowmick (US (US 2009 / 0017463), Srivastava et al. (US 2008 / 0269157), Chinnaiyan et al. (US 2008 / 0222741), Thaxton et al. (US 2008 / 0181850), Dahary et al. (US 2008 / 0014590), Diamandis et al. (US 2006 / 0269971), Rubin et al. (US 2006 / 0234259), Einstein et al. (US 2006 / 0115821), Paris et al. (US 2006 / 0110759), Condon-Cardo (US 2004 / 0053247), and Ritchie et al. (US 2009 / 0127454). Biomarkers associated with the development of pancreatic cancer have been described by Sahin et al. (US 7,527,933), Rataino et al. (US 7,507,541), Schleyer et al. (US 7,476,506), Domon et al. (US 7,473,531), McCaffey et al. (US 7,358,231), Price et al. (US 7,229,770), Chan et al. (US 2005 / 0095611), Mitchl et al. (US 2006 / 0258841), and Faca et al. (PLoS Med 5(6):e123, 2008).Biomarkers associated with the development of lung cancer have been reported by Sahin et al. (US 7,527,933), Hutteman (US 7,473,530), Bae et al. (US 7,368,255), Wang (US 7,348,142), Nacht et al. (US 7,332,590), Gure et al. (US 7,314,721), Patel (US 7,300,765), Price et al. (US 7,229,770), O'Brien et al. (US 7,291,462 and US 6,316,213), Muramatsu et al. (US 7,090,983), Carson et al. (US 6,576,420), Giordano (US 5,840,506), Guo (US (US 2009 / 0062144), Tsao et al. (US 2008 / 0176236), Nakamura et al. (US 2008 / 0050378), Raponi et al. (US 2006 / 0252057), Yip et al. (US 2006 / 0223127), Pollock et al. (US 2006 / 0046257), Moon et al. (US 2003 / 0224509), and Budiman et al. (US 2009 / 0098543). Biomarkers associated with the development of skin cancer (e.g., basal cell carcinoma, squamous cell carcinoma, and melanoma) have been described by Roberts et al. (US 6,316,208), Polsky (US 7,442,507), Price et al. (US 7,229,770), Genetta (US 7,078,180), Carson et al. (US 6,576,420), Moses et al. (US 2008 / 0286811), Moses et al. (US 2008 / 0268473), Dooley et al. (US 2003 / 0232356), Chang et al. (US 2008 / 0274908), Alani et al. (US 2008 / 0118462), Wang (US 2007 / 0154889), and Zetter et al. (US This is shown in the attached document (No. 2008 / 0064047).Biomarkers associated with the development of multiple myeloma have been described in Coignet (US 7,449,303), Shaughnessy et al. (US 7,308,364), Seshi (US 7,049,072), and Shaughnessy et al. (US 2008 / 0293578, US 2008 / 0234139, and US 2008 / 0234138). Biomarkers associated with the development of leukemia have been described by Ando et al. (US 7,479,371), Coignet (US 7,479,370 and US 7,449,303), Davi et al. (US 7,416,851), Chiorazzi (US 7,316,906), Seshi (US 7,049,072), Van Baren et al. (US 6,130,052), Taniguchi (US 5,643,729), Insel et al. (US 2009 / 0131353), and Van Bockstaele et al. (Blood Rev. 23(1):25-47, 2009). Biomarkers associated with the development of lymphoma are provided in Ando et al. (US 7,479,371), Levy et al. (US 7,332,280), and Arnold (US 5,858,655). Biomarkers associated with the development of bladder cancer are provided in Price et al. (US 7,229,770), Orntoft (US 6,936,417), Haak-Frendscho et al. (US 6,008,003), Feinstein et al. (US 6,998,232), Elting et al. (US 2008 / 0311604), and Wewer et al. (2009 / 0029372). The contents of each of the above references are incorporated herein by reference in their entirety.

[0156] In certain embodiments, the method of the present invention can be used to monitor a patient for cancer recurrence. The patient has already been treated for cancer, so the genetic profile and the specific mutation(s) associated with the patient's cancer are already known. A probe can be designed that specifically hybridizes to the region of nucleic acid that contains the mutation(s) that are indicative of the cancer for which the patient has previously been treated. The patient's sample (e.g., pus, sputum, semen, urine, blood, saliva, feces, or cerebrospinal fluid) can then be analyzed as described above to determine whether mutant allele(s) are detected in the sample, which mutant allele(s) are indicative of cancer recurrence.

[0157] Droplet sorting The methods of the invention include sorting of droplets based on whether they contain a homogeneous or heterogeneous population of molecules. A sorting module may be a junction of channels where the flow of droplets can be redirected into one or more other channels, e.g., branch channels, depending on a signal received in connection with interrogation of the droplets in a detection module. Typically, the sorting module is monitored and / or under the control of the detection module, and thus can be responsive to the detection module. The sorting region is in communication with and influenced by one or more sorting devices.

[0158] The sorting device includes techniques or control systems, e.g., dielectric, electrical, electroosmotic, (micro-)valves, etc. The control system can use various sorting techniques to redirect or direct the flow of molecules, cells, small molecules or particles into a given branch channel. A branch channel is a channel that communicates with a sorting region and a main channel. The main channel may communicate with two or more branch channels at a sorting module or branch point, for example to form a T-shape or a Y-shape. Other shapes and channel geometries can be used as desired. Typically, the branch channel receives droplets of interest once they have been detected by the detection module and sorted in the sorting module. The branch channel can have an outflow module and / or can have a well or reservoir at the end that allows for the collection or disposal of the molecules, cells, small molecules or particles (collection module or disposal module, respectively). Alternatively, the branch channel may communicate with other channels to allow additional sorting.

[0159] A property of the fluidic droplet can be sensed and / or determined in some manner, for example as described herein (e.g., the fluorescence of the fluidic droplet can be determined), and in response, an electric field can be applied or removed to direct the fluidic droplet to a particular region (e.g., a channel). In certain embodiments, the fluidic droplet is sorted or directed by inducing a dipole in an uncharged fluid droplet (which may be initially charged or uncharged) and sorting or directing the droplet with an applied electric field. The electric field can be an AC field, a DC field, or the like. For example, a channel containing the fluidic droplet and the carrier fluid splits into a first and a second channel at a branch point. Generally, the fluidic droplet is uncharged. After the branch point, a first electrode is positioned near the first channel and a second electrode is positioned near the second channel. A third electrode is positioned near the branch point of the first and second channels. A combination of these electrodes is then used to induce a dipole in the fluidic droplet. Thus, by applying an appropriate electric field, the droplets can be directed into either the first or second channel as desired. Further description of droplet sorting is provided, for example, in Link et al. (U.S. Patent Applications 2008 / 0014589, 2008 / 0003142, and 2010 / 0137163) and in Raindance Technologies Inc.'s European Patent Application Publication No. 2047910.

[0160] Based on the signal detected by the detection module, the droplets containing a heterogeneous population of molecules are sorted out from the droplets containing a homogeneous population of molecules. The droplets can be further sorted to separate the droplets containing a homogeneous population of target amplicons from the droplets containing a homogeneous population of target mutant amplicons.

[0161] Release of targets from droplets The method of the present invention can further comprise releasing the amplified target molecules from the droplet for further analysis.Methods of releasing the amplified target molecules from the droplet are shown, for example, in Link et al. (U.S. Patent Application Nos. 2008 / 0014589, 2008 / 0003142 and 2010 / 0137163) and in Raindance Technologies Inc.'s European Patent Application Publication No. 2047910.

[0162] In certain embodiments, the sample droplets are creamed onto the carrier fluid. As a non-limiting example, the carrier fluid can be a perfluorocarbon oil that can have one or more stabilizing surfactants. The droplets rise or separate from the carrier fluid due to the density of the carrier fluid being greater than that of the aqueous phase that constitutes the droplet. For example, the perfluorocarbon oil used in one embodiment of the method of the present invention has a density of 1.8 compared to the density of the aqueous phase of the droplet, which is 1.0.

[0163] The creamy liquids are then placed onto a second carrier fluid containing a destabilizing surfactant, such as a perfluorinated alcohol (e.g., 1H,1H,2H,2H-perfluoro-1-octanol). The second carrier fluid can also be a perfluorocarbon oil. Upon mixing, the aqueous droplets begin to coalesce, and the coalescence is completed by a short centrifugation at low speed (e.g., 1 minute at 2000 rpm in a centrifuge). The coalesced aqueous phase can now be removed and further analyzed.

[0164] The released amplified material can also be subjected to further amplification by using tailed primers and second PCR primers. In this embodiment, the primers in the droplets contain additional sequences or tails added to the 5' end of the sequence-specific portion of the primers. The sequences for these tailed regions are the same for each primer pair and are incorporated into the 5' portion of the amplicon during PCR cycling. Once the amplicons are removed from the droplets, the products can be further amplified by additional PCR rounds using another set of PCR primers that can hybridize to the tailed region of the amplicon. The second primers may be exactly matched in length and sequence to the tailed region, or may themselves contain additional sequences at the 5' end of the tail portion of the primers. During the second PCR cycling, these additional regions also become incorporated into the amplicon. These additional sequences can include, but are not limited to, adapter regions used by the sequencing platform for library preparation and sequencing, sequences used as barcoding functions for identification of samples multiplexed into the same reaction, molecules for separation of the amplicon from the rest of the reaction materials such as biotin, digoxin, peptides or antibodies, and molecules such as fluorescent markers that can be used to identify fragments.

[0165] In certain embodiments, the amplified target molecules are sequenced.In certain embodiments, the sequencing is single-molecule sequencing-by-synthesis.Single-molecule sequencing is described, for example, in Lapidus et al. (U.S. Pat. No. 7,169,560), Quake et al. (U.S. Pat. No. 6,818,395), Harris (U.S. Pat. No. 7,282,337), Quake et al. (U.S. Pat. App. No. 2002 / 0164629), and Braslavsky et al., PNAS (USA), 100:3960-3964 (2003), the contents of each of these references being incorporated herein by reference in their entirety.

[0166] Briefly, single-stranded nucleic acids (e.g., DNA or cDNA) are hybridized to oligonucleotides attached to the surface of a flow cell. The single-stranded nucleic acids can be captured by methods known in the art, such as those shown in Lapidus (US Pat. No. 7,666,593). The oligonucleotides can be covalently attached to the surface, or various non-covalent attachments can be used, as known to those skilled in the art. Furthermore, the attachment can be indirect, such as by the polymerase of the present invention attached directly or indirectly to the surface. The surface can be flat or not, and / or porous or non-porous, or any other type known to those skilled in the art to be suitable for attachment. The nucleic acids are then sequenced by imaging the polymerase-mediated addition of fluorescently labeled nucleotides incorporated into the growing strand surface oligonucleotides at single molecule resolution. EXAMPLES

[0167] Experimental details What follows next are experimental details for the various experiments detailed above.

[0168] Primers and probes All TaqMan® primers and probes used here are listed in Table 2. Unless otherwise noted by reference in the table, primers and probes were designed with the "Custom TaqMan® Assay Design Tool" from Applied Biosystems Inc. (ABI) and procured through ABI (Carlsbad, Calif.). Probes were designed with 6-carboxyfluorescein (FAM, λ 励起 494nm\λ 放射 494 nm) or VIC™ (from ABI, λ 励起 538nm\lambda 放射 The antibody was labeled with a 554 nm fluorescence intensity.

[0169] [Table 2] Target DNA For some gene targets, BCKDHA and SMN2, plasmid DNA containing sequences spanning between the primer pairs (see Table 2) was synthesized (GeneArt, Regensburg, Germany) and cloned into a GeneArt standard vector (2.5 kb). To avoid any DNA supercoiling that may affect the assay, the target fragments were released from the cloning vector by restriction digestion with SfiI. For simplicity, these gene fragments are referred to throughout the text as "plasmid DNA". For demonstration experiments of multiplex reactions, strings of different gene fragments were also synthesized (GeneArt) and cloned into a GeneArt standard vector, which are referred to in the text as "artificial chromosomes". In this case, the fragments were separated from each other by restriction digestion at adjacent EcoRV sites. Human DNA was obtained in already purified form from cell lines (see Table 3; Coriell, Camden, NJ) and fragmented with a K7025-05 nebulizer according to the manufacturer's instructions (Invitrogen, Carlsbad, CA) before use. DNA concentration was quantified by measuring absorbance at 260 nm using a Nanodrop 2000 spectrophotometer (Thermo Scientific, Wilmington, DE).

[0170] [Table 3] Microfluidics Microfluidic chips were fabricated by conventional soft lithography. A molding master was created by spin-coating SU-8 negative photoresist (MicroChem Corp., Newton, MA) onto a 6-inch silicon wafer and transferring fluidic geometries from a photomask (CAD / Art Services, Bandon, OR) by contact lithography on an OAI Hybralign Series 200 aligner (OAI, San Jose, CA). The chip contained channels with two depths: deep channels (100±10 um) with low hydrodynamic resistance for transporting fluids from the chip's external ports to the functional areas, and shallow channels (20±1 um) for droplet manipulation and detection. SU-8 photoresists 2100 and 2025 were used for the deep and shallow channels, respectively. Polydimethylsiloxane (PDMA) (Sylgard® 184, Dow Corning, Midland, MI) chips were molded from the negative master in custom-designed mold housings. Glass cover slides were permanently attached to the fluidic side of the chips by surface activation with an AutoGlow™ oxygen plasma system (Glow Research, Phoenix, AZ) followed by immediate pressure bonding. To create a hydrophobic surface, the microfluidic channels were exposed for ∼2 min to 1H,1H,2H,2H-perfluorodecyltrichlorosilane (Alfa Aesar, Ward Hill, MA) dissolved in FC-3283 (3M Specialty Materials, St. Paul, MN), prepared as a mixture of 18 g of silane in 100 uL of solvent. Two different microfluidic devices were used, one for droplet generation and the other for fluorescent readout after thermal cycling. The droplet generation chip gave rise to an emulsion of uniformly sized aqueous droplets of template DNA and PCR master mix suspended in an inert fluorinated oil with an emulsion-stabilizing surfactant (REB carrier oil; RainDance Technologies, Lexington, MA), referred to from this point on as “carrier oil.” Droplets were generated within a cross-shaped microfluidic intersection, or "nozzle."As shown in Figure 3a, under typical operation, the aqueous phase flowed into the nozzle from the right (160 uL / hr) and merged with carrier oil flows from the top and bottom (total oil at 750 uL / hr), generating 4 pL droplets at a rate of 11 kHz. Channel widths at the intersection measured 15 um for the aqueous inlet, 12.5 for the oil inlet, and 15 um widening to 40 um at the outlet. Flow was driven by a custom OEM pump (IDEX Corporation, Northbrook, IL).

[0171] Approximately 25 uL of PCR reaction mixture was collected from the droplet generation chip as an emulsion and subjected to thermal cycling on a DNA Engine (Bio-Rad, Hercules, CA). The reaction mixture contained 1× TaqMan® Universal PCR Master Mix (Applied Biosystems, Carlsbad, CA), 0.2 mM dNTPs (Takara Bio, Madison, WI) and varying amounts of primer pairs and probes as described in the results. In all cases, the same amounts of primers and probes were varied when varying from the 1× concentration. The cycle program included a 10 min hot start at 95° C. and 45 cycles of 15 s at 95° C. and 60 s at 60° C.

[0172] The droplets had become concentrated during off-chip processing as the carrier oil became more concentrated than the aqueous phase and flowed out of the emulsion. The droplets were therefore re-injected into the readout chip as a tightly packed emulsion that required dilution before readout to properly distinguish one droplet from the other. Using a "spacer" nozzle similar to the droplet-generating nozzle above, a uniform plug of excess carrier oil was injected between the droplets just prior to readout. As shown in Figure 3b, the droplet inlet to the nozzle tapered to a constriction of approximately the size of an individual droplet, forcing the droplets to enter the nozzle in a single file and therefore at a steady rate. Countercurrent flows of carrier oil from the upper and lower channels separated them uniformly. The nozzle exiting the spacer nozzle increased in width along the flow direction, and the droplets were interrogated by laser-induced fluorescence at positions along that channel (marked by arrows in Figure 3b) where their width was less than or equal to the droplet diameter. The nozzle dimensions were 15um for the droplet inlet and outlet, and 20um for the oil line.

[0173] instrument use Fluorescence readout was performed by conventional epifluorescence microscopy on a custom-built microscope. A 20 mW, 488 nm laser source (Cyan; Picarro, Sunnyvale, CA) was magnified 2× and focused onto the microfluidic channel by an objective lens (20× / 0.45NA; Nikon, Japan). Two band-pass filters discriminated the fluorescence collected by the objective lens: 512 / 25 nm and 529 / 28 nm for FAM and VIC fluorophores, respectively (Semrock, Rochester, NY). Fluorescence was detected by two H5784-20 photomultiplier tubes (Hamamatsu, Japan) and typically recorded at a sampling rate of 200 kHz using a USB-6259 data acquisition card (National Instruments, Austin, TX). Data traces were smoothed by a seven-point quadratic Savitzky-Golay algorithm before subsequent analysis. Concurrent with the fluorescence readout was an 850 nm LED (TSHG6200; Vishay, Shelton, CT). Droplets were imaged with the same objective lens with back illumination from Sigma Semiconductors, a short-pass filter to separate the light paths for fluorescence detection and imaging, and a Guppy CCD camera (Allied Vision Technologies, Newburyport, MA). Droplets were imaged with short illumination pulses (5–20 μs) to avoid streaking in the images.

[0174] Data analysis Data were analyzed with custom LabView software (National Instruments, Austin, TX), which interpreted droplet events as sustained bursts of fluorescence intensity above a threshold. Signal-to-noise ratios were generally fairly high and signal levels were consistent from day to day, therefore a fixed threshold of 50 mV was primarily used, whereas in other situations the threshold was set visually. Peak fluorescence intensity was recorded for each droplet event for both VIC and FAM fluorophores. Some coalescence of droplets occurred during thermal cycling, generally as isolated events between two intact droplets to form "doublets". Doublets and rare larger coalescence events were easily filtered from the dataset based on the duration of the fluorescence burst.

[0175] Incorporation by Reference References and citations to other documents, such as patents, patent applications, patent publications, journals, books, articles, web content, etc., are made throughout this disclosure. All such documents are incorporated herein by reference in their entirety for all purposes.

[0176] Equivalent The present invention may be embodied in other specific forms without departing from its spirit or essential characteristics. The above-described embodiments are therefore to be considered in all respects illustrative and not restrictive of the invention described herein.

Claims

1. A microdroplet comprising a single nucleic acid template and multiple primer pairs specific for multiple target sites on the template.

2. The microdroplet of claim 1 , further comprising a plurality of probes that hybridize to amplicons produced within the droplet.

3. The microdroplet of claim 1 , wherein the single nucleic acid template is DNA or RNA.

4. The microdroplet of claim 1 further comprising reagents for performing a polymerase chain reaction.

5. The microdroplet of claim 2 , wherein a member of the plurality of probes contains a detectable label.

6. The microdroplet of claim 6 , wherein the plurality of probes comprises one or more groups of probes of varying concentrations.

7. The microdroplet of claim 6 , wherein each member of the one or more groups of probes comprises the same detectable label.

8. The microdroplet of claim 5 , wherein each member of the plurality of probes comprises a different detectable label.

9. The microdroplet of claim 5 , wherein the detectable label is a fluorescent label.

10. 1. A method for detecting multiple targets in a biological sample, comprising: a) forming one or more microdroplets, each containing a single nucleic acid template and a heterogeneous mixture of primer pairs and probes, each specific for multiple target sites on the template; b) amplifying the nucleic acid template in the one or more microdroplets; and c) analyzing said one or more microdroplets The method includes:

11. The forming step comprises: a) providing a first fluid comprising a single nucleic acid template; b) providing a second fluid comprising a plurality of primers and a plurality of probes, each specific for a plurality of target sites on said template; c) combining the first fluid with the second fluid to form droplets containing the single nucleic acid template and a heterogeneous mixture of primer pairs and probes. The method of claim 10, comprising:

12. The method of claim 11 , wherein the first fluid and the second fluid are combined in the presence of an electric field relative to the fluids.

13. The method of claim 11 , wherein the first fluid and the second fluid are microdroplets.

14. The method of claim 11 , wherein the second fluid further comprises reagents for performing a polymerase chain reaction.

15. The method of claim 10, wherein the single nucleic acid template is DNA or RNA.

16. The method of claim 10 , wherein the analyzing step comprises detecting the presence or absence of the plurality of targets within the one or more microdroplets.

17. The method of claim 10 , wherein a member of the plurality of probes contains a detectable label.

18. 20. The microdroplet of claim 17, wherein the plurality of probes comprises one or more groups of probes of varying concentrations.

19. 20. The microdroplet of claim 18, wherein each member of the one or more groups of probes comprises the same detectable label.

20. 20. The microdroplet of claim 17, wherein each member of the plurality of probes comprises a different detectable label.

21. The microdroplet of claim 17 , wherein the detectable label is a fluorescent label.

22. 1. A method for analyzing a target nucleic acid, comprising: forming a droplet containing a single target nucleic acid and one or more amplification reagents; amplifying the target within the droplet; Displacing droplets containing amplicons from the target and amplicons from variants of the target; and Analyzing the target amplicons The method includes:

23. 23. The method of claim 22, wherein the amplifying step is a polymerase chain reaction and the one or more amplification reagents comprise one or more primer pairs.

24. 23. The method of claim 22, wherein the removing step comprises flowing the droplet into a microfluidic channel.

25. 23. The method of claim 22, wherein the analyzing step comprises detecting the amplicons by hybridization to a detectably labeled probe.

26. 23. The method of claim 22, wherein the analyzing step is performed on amplicons from droplets that were not excluded in the exclusion step.

27. The forming step comprises: flowing a stream of a first sample fluid containing nucleic acids across two opposing streams of a flowing carrier fluid, thereby forming a plurality of first droplets containing said first sample fluid, said carrier fluid being immiscible with said sample fluid; and combining each of a plurality of first droplets comprising the first sample fluid with a portion of a second fluid comprising one or more amplification reagents, the portion of the second fluid being optionally a droplet; 23. The method of claim 22, comprising:

28. 28. The method of claim 27, wherein the carrier fluid is an oil.

29. 30. The method of claim 28, wherein the oil comprises a surfactant.

30. 30. The method of claim 29, wherein the surfactant is a fluorosurfactant.

31. The analyzing step comprises: Determining the number of droplets containing only wild type targets; Determining the number of droplets containing only the target variant.

23. The method of claim 22, comprising:

32. 32. The method of claim 31 , wherein the presence of droplets containing only the mutant is indicative of disease.

33. 33. The method of claim 32, wherein the disease is cancer.

34. 32. The method of claim 31 , wherein the variant is an allelic variant.

35. 35. The method of claim 34, wherein the allelic variant is a single nucleotide polymorphism.

36. 1. A method for processing a nucleic acid sample, comprising: obtaining a droplet containing one nucleic acid target; amplifying the nucleic acid in the droplet; and Separating the droplets comprising the heterogeneous population of amplicons from the droplets containing the homogeneous population of amplicons. The method includes:

37. 1. A method for detecting recurrence of cancer in a patient, comprising: forming sample droplets, each droplet containing, on average, a single target nucleic acid taken from a patient sample; flowing the sample droplet through a channel; amplifying targets within the droplets; detecting the amplified target within the droplet; rejecting droplets containing a heterogeneous population of amplicons; and Analyzing the non-excluded droplets to determine the presence of mutant alleles indicative of recurrence. The method includes:

38. The forming step comprises: Flowing a stream of sample fluid containing nucleic acids perpendicularly through two opposing streams of flowing carrier fluid, said carrier fluid being immiscible with said sample fluid.

38. The method of claim 37, comprising:

39. 38. The method of claim 37, wherein the sample is human tissue or body fluid.

40. 40. The method of claim 39, wherein the bodily fluid is selected from the group consisting of pus, sputum, semen, urine, blood, saliva and cerebrospinal fluid.

41. 38. The method of claim 37, wherein the analyzing step comprises capturing amplicons obtained from the droplets using a labeled capture probe.

42. 1. A method for analyzing a target nucleic acid, comprising: compartmentalizing the first fluid into a plurality of portions, each portion containing a single target nucleic acid; amplifying targets in said portions; eliminating portions containing amplicons from the target and amplicons from variants of the target; and Analyzing the target amplicons A method comprising:

43. 43. The method of claim 42, wherein the compartment comprises a chamber in a microfluidic device.

44. 1. A method for analyzing a target nucleic acid, comprising: forming a sample droplet containing a single target nucleic acid; amplifying targets within the droplets; Displacing droplets containing amplicons from the target and amplicons from variants of the target; and Analyzing the target amplicons A method comprising:

45. 12. The method of claim 11 , wherein the first fluid is contained within a first droplet and the second fluid is a first portion of a stream, and further wherein step (c) comprises combining the first droplet with the first portion.

46. 12. The method of claim 11 , wherein the second fluid is contained within a first droplet and the first fluid is a first portion of a stream, and further wherein step (c) comprises combining the first droplet with the first portion.

Images