Use of unique molecular identifiers to improve accuracy of long read sequencing and characterization of CRISPR edits

Abstract

Described herein is a system and method for long read sequencing that generates a single molecule consensus sequence for each starting molecule in a population of samples using PCR primers incorporating unique molecule identifiers (UMIs). This method reduces the sequencing error rate by generating consensus sequences from individual reads in each UMI group, balancing sequencing errors to improve confidence in actual sequences, thereby allowing for increased accuracy in quantifying accurate knock-in events, and reporting perfect HDR integration.

Description

[0001] CROSS-REFERENCE TO RELATED APPLICATIONS

[0002] This application claims priority to U.S. Provisional Patent Application No. 63 / 341,850, filed May 13, 2022, which is incorporated herein by reference in its entirety. This application is related to U.S. Patent Application No. 16 / 919,577 and International Patent Application No. PCT / US2020 / 040621, both filed on July 2, 2020, each of which is incorporated herein by reference in its entirety.

[0003] References to sequence listings

[0004] In accordance with 37 CFR § 1.831 and PCT Rule 13ter, this application is filed with a sequence listing XML in ST.26 XML format. The sequence listing XML file "013670-9069-WO01_sequence_listing_xml_28-APR-2023.xml" filed at the USPTO Patent Center was created on April 28, 2023, contains 26 sequences, has a file size of 48.8Kb, and is incorporated by reference into this specification as a whole. Technical Field

[0005] This article describes a system and method for long-read sequencing that uses polymerase chain reaction (PCR) primers that incorporate unique molecular identifiers (UMIs) to generate single-molecule consensus sequences for each starting molecule in a sample population. This method reduces the sequencing error rate by generating a consensus sequence from individual reads in each UMI group, balancing sequencing errors to increase confidence in the actual sequence, thereby allowing increased accuracy in quantifying accurate knock-in events, and reporting perfect homology-directed repair (HDR) integration. Background Art

[0006] Long-read sequencing using Oxford Nanopore Technologies (ONT) and / or PacBio sequencing platforms is advantageous over existing methods in characterizing CRISPR-Cas9 homology-directed repair (HDR)-mediated knock-ins because it provides phased information of the entire edited genomic locus, including the knock-in sequence. This allows confirmation that the exogenous sequence of interest has been integrated as expected. However, long-read sequencing using currently available platforms is very prone to sequencing errors, which limits the utility of these systems in obtaining accurate bases by base parsing the knock-in sequence in a highly diverse polyclonal population. There is often not enough edited genomic DNA to use a non-amplification enrichment strategy (e.g., target enrichment), so PCR is required to generate enough material for sequencing. However, during library preparation, PCR amplification of genomic DNA samples may result in a biased presentation of wild-type (WT) and HDR sequences in the final sequencing library because there are differences in the amplification efficiency of shorter WT sequences and longer knock-in sequences. This "PCR bias" may artificially reduce the measured HDR frequency, resulting in failure to represent the actual knock-in integration efficiency.

[0007] Algorithms and methods (together forming new methods) that improve the accuracy of long-read sequencing and the characterization of CRISPR editing are needed. These methods will be useful for characterizing the performance of CRISPR-Cas9 HDR-mediated knock-in applications. Summary of the invention

[0008] One embodiment described herein is a method for improving the accuracy of long-read sequencing, the method comprising: generating a sequencing library, comprising: (a) amplifying a locus with primers comprising a unique molecular identifier and a universal sequence to generate an initial product; (b) purifying the initial product; (c) amplifying the initial product with primers comprising a sequence complementary to the universal sequence and a barcode sequence to generate a barcoded product; (d) purifying the barcoded product to generate a purified barcoded product; (e) pooling the purified barcoded products to generate pooled barcoded products; and (f) sequencing the pooled barcoded products using a long-read sequencing device to generate raw nucleotide sequence data. In one aspect, the method further comprises executing on a processor: (g) receiving raw nucleotide sequence data; (h) aligning the raw nucleotide sequence data with a reference amplicon to generate a mapping sequence; (i) identifying and separating mapping sequences by target region to generate multiple groups of target region sequences; (j) for each group of target region sequences: (i) analyzing the unique molecular identifiers of the target region sequences and discarding target region sequences lacking unique molecular identifiers; (ii) clustering the target region sequences containing unique molecular identifiers to generate clustered target region sequences and cluster consensus sequences; (iii) analyzing and filtering the clustered target region sequences, discarding sequences with a cluster consensus sequence number less than a selected number, and downsampling clusters larger than a selected cluster size to a selected cluster size; (iv) generating an initial target sequence consensus sequence; (k) repeating step (j) for the initial target sequence consensus sequence to generate a high-accuracy consensus sequence for each cluster group, and correcting amplification bias by clustering groups that were not similar enough to be clustered in the first round; and (l) outputting high-accuracy consensus sequence data. In another aspect, step (j)(i) comprises: aligning the 5' and 3' adapters and UMI-adjacent substrings of the target region with the two terminal strings of the sequence; identifying the nucleotides between the aligned target sequence and adapter sequence at each end and clustering the UMI sequences; and discarding sequences that lack UMIs at both ends and have an edit difference of less than 3 with the UMIs. In another aspect, the number of selected cluster consensus sequences is between 3 and 10; and the selected cluster size is 20 to 80.In another aspect, the method further comprises analyzing the raw nucleotide sequence data of step 1(f) or the high-accuracy consensus sequence data of step 2(l), comprising executing on a processor: receiving sequence data comprising a plurality of sequences; analyzing and merging the sample sequence data, and outputting the merged sequence; when a single-stranded or double-stranded DNA oligonucleotide donor is provided, developing a target site sequence containing a predicted result of a repair event and outputting a target prediction result; using a mapper, binning the merged sequence with the target site sequence or an optional target prediction result, and outputting a target read alignment; using an enzyme-specific position-specific scoring matrix derived from biological data based on the position of a guide sequence and a typical enzyme-specific cleavage site, re-aligning the binned target read alignment with the target site, and generating a final alignment; analyzing the final alignment within a predefined sequence distance window and identifying and quantifying mutations from the typical enzyme-specific cleavage site; outputting the final alignment, analysis and quantification result data as a table or graph. In another aspect, the purification in steps (b) and (d) comprises solid phase reversible immobilization (SPRI) purification. In another aspect, the unique molecular identifier comprises 8 to 30 nucleotides. In another aspect, the unique molecular identifier comprises 8 to 18 nucleotides. In another aspect, the universal sequence comprises 22 to 30 nucleotides. In another aspect, the barcode sequence comprises 16 to 24 nucleotides. In another aspect, the amplification in step (a) comprises at least 2 PCR cycles. In another aspect, the amplification in step (a) comprises 2 to 4 PCR cycles. In another aspect, the amplification in step (c) comprises 20 to 40 PCR cycles. In another aspect, the long read sequencing device is selected from Oxford Nanopore Technologies (ONT) MinION or PacBio Sequel II. In another aspect, the sequencing error rate is reduced by at least 15 times.

[0009] Description of the drawings

[0010] Figure 1 The library preparation workflow is shown. Using two PCR cycles with primers containing UMIs, unique UMIs are incorporated on each end of each input DNA molecule. The universal tail is then used to amplify the reaction, generating multiple copies of each UMI set for sequencing.

[0011] Figure 2 An improved UMI consensus construction workflow from Oxford Nanopore Technologies' github is shown.

[0012] FIG. 3A to FIG. 3B The fraction of HDR reads considered perfect in the CRISPAltRations output without UMI consensus construction (original) or with UMI consensus construction using min10 or min3 parameters is shown. Figure 3A Shown are the results of sequencing using the Oxford Nanopore Technologies MinION device (R9.4.1 chemistry, Kit9). Figure 3B Results from sequencing on a PacBio Sequel II apparatus are shown (HiFi CCS reads, min20Q score).

[0013] Figure 4 Shown are CRISPAltRations HDR quantifications without UMI consensus construction (original) or using UMI consensus construction with a UMI sequence similarity of 80% and a minimum intermediate cluster size of 3 or 10. The expected HDR percentages for the samples are plotted with repixelation.

[0014] Figure 5 The general workflow of CRISPAltRations for short and long read sequences is shown. The edited genomic DNA is extracted and amplified using targeted multiplex PCR to enrich for on-target loci and predicted off-target loci. The amplicons are sequenced on an Illumina MiSeq. Read pairs are merged into single fragments (FLASH), mapped to the genome (minimap2), and binned according to their alignment with the expected amplicon position. After finding the cut site and creating a position-specific gap opening / extension reward matrix, the reads in each bin are realigned with the expected amplicon sequence to preferentially align the indel / expected indel profiles closer to the cut site for each enzyme (CRISPAltRations encoding + psnw). Indels that intersect with windows upstream or downstream of the cut site are annotated. The editing percentage is the sum of reads containing indels / total number of observations. DETAILED DESCRIPTION

[0015] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by those of ordinary skill in the art. For example, any nomenclature and techniques used in conjunction with the biochemistry, molecular biology, immunology, microbiology, genetics, cell and tissue culture, and protein and nucleic acid chemistry described herein are well known and commonly used in the art. In the event of a conflict, the present disclosure, including definitions, will prevail. Exemplary methods and materials are described below, but methods and materials similar or equivalent to those described herein may be used in the practice or testing of the embodiments and aspects described herein.

[0016] As used herein, the terms "amino acid", "nucleotide", "polynucleotide", "vector", "polypeptide" and "protein" have the ordinary meaning understood by ordinary biochemists in the art. Standard single-letter nucleotides (A, C, G, T, U) and standard single-letter amino acids (A, C, D, E, F, G, H, I, K, L, M, N, P, Q, R, S, T, V, W or Y) are used herein. As used herein in nucleotide sequences, "N" refers to any nucleotide, such as A, T, C, G; "R" refers to a purine nucleotide, such as C or G; "Y" refers to a pyrimidine nucleotide, such as A or T. Some nucleotide sequences have a 5'-amino-C6 modification, for example, 5'-NH2(CH2)6PO4-, abbreviated as " / 5AmMC6 / ".

[0017] As used herein, terms such as "including," "comprising," "having," and the like mean "comprising." The present disclosure also contemplates other embodiments that "comprising," "consisting of," and "consisting essentially of" the embodiments or elements set forth herein, whether explicitly stated or not.

[0018] Unless otherwise indicated herein or clearly contradicted by context, the terms "a", "an", "the" and similar terms used in the context of this disclosure (especially in the context of the claims) as used herein will be deemed to cover both the singular and the plural. Furthermore, unless otherwise specified, "a", "an" or "the" means "one (kind) or more (kinds)".

[0019] The term "or" as used herein may be a conjunction or a disjunctive conjunction.

[0020] As used herein, the term "substantially" means to a great or significant extent, but not completely.

[0021] As used herein, the term "about" or "approximately" refers to a value similar to the stated reference value or within an acceptable error range determined by a person of ordinary skill in the art for a particular value when applied to one or more target values, and the error range will depend in part on how to measure or determine the value, such as the limitations of the measurement system. In one aspect, the term "about" refers to any value within a variation range of up to ±10% of the value modified by the term "about", including both integer and fractional components. Alternatively, according to the practice of the art, "about" can mean within 3 times or more of the standard deviation. Alternatively, as for a biological system or process, the term "about" can mean within an order of magnitude of the value, within 5 times in some embodiments, and within 2 times in some embodiments. As used herein, the symbol "~" means "about" or "approximately".

[0022] All ranges disclosed herein include both endpoints as discrete values ​​and all integers and fractions specified within the range. For example, the range of 0.1-2.0 includes 0.1, 0.2, 0.3, 0.4 ... 2.0. If the endpoint is modified by the term "about", the specified range will expand the range of variation (including the endpoint) within ±10% or more than 3 times the standard deviation of any value within the range.

[0023] As used herein, the terms "control" or "reference" are used interchangeably herein. A "reference" or "control" level can be a predetermined value or range used as a baseline or benchmark for evaluating measurement results. "Control" also refers to a control experiment.

[0024] Described herein is the use of unique molecular identifiers (UMIs) incorporated within PCR primers to allow generation of single-molecule consensus sequences for each starting molecule in a sample population. This approach can correct for PCR bias and allow more accurate counting of the number of starting molecules prior to PCR amplification. By extension, for the present application, merging reads by matching UMIs enables better quantification of HDR frequencies. In addition, this approach balances sequencing errors to increase confidence in the actual sequence by generating a consensus sequence from individual reads in each UMI group, thereby allowing increased accuracy in quantifying accurate knock-in events and reporting perfect HDR integration to reduce sequencing error rates.

[0025] In this approach, PCR primers are designed to include target-specific sequences, UMIs, and universal 5' ends for a secondary barcoding step, allowing for sample multiplexing during sequencing runs (see Figure 1 ). After sequencing, base calling, and demultiplexing, multiple sequencing reads for each UMI are used to construct a consensus sequence through a bioinformatics pipeline. The final UMI consensus sequence can be used for downstream analysis, such as CRISPR editing characterization and HDR quantification using the CRISPAltRations analysis pipeline (Synthetic DNA Technologies), as described in U.S. Patent Application Publication No. US 2021 / 0002700 A1 and International Patent Application Publication No. WO 2021 / 003343 A1, the teachings of which are each incorporated herein by reference.

[0026] One embodiment described herein is a method for improving the accuracy of long-read sequencing, the method comprising: generating a sequencing library, comprising: (a) amplifying a locus with primers comprising a unique molecular identifier and a universal sequence to generate an initial product; (b) purifying the initial product; (c) amplifying the initial product with primers comprising a sequence complementary to the universal sequence and a barcode sequence to generate a barcoded product; (d) purifying the barcoded product to generate a purified barcoded product; (e) pooling the purified barcoded products to generate pooled barcoded products; and (f) sequencing the pooled barcoded products using a long-read sequencing device to generate raw nucleotide sequence data. In one aspect, the method further comprises executing on a processor: (g) receiving raw nucleotide sequence data; (h) aligning the raw nucleotide sequence data with a reference amplicon to generate a mapping sequence; (i) identifying and separating mapping sequences by target region to generate multiple groups of target region sequences; (j) for each group of target region sequences: (i) analyzing the unique molecular identifiers of the target region sequences and discarding target region sequences lacking unique molecular identifiers; (ii) clustering the target region sequences containing unique molecular identifiers to generate clustered target region sequences and cluster consensus sequences; (iii) analyzing and filtering the clustered target region sequences, discarding sequences with a cluster consensus sequence number less than a selected number, and downsampling clusters larger than a selected cluster size to a selected cluster size; (iv) generating an initial target sequence consensus sequence; (k) repeating step (j) for the initial target sequence consensus sequence to generate a high-accuracy consensus sequence for each cluster group, and correcting amplification bias by clustering groups that were not similar enough to be clustered in the first round; and (l) outputting high-accuracy consensus sequence data. In another aspect, step (j)(i) comprises: aligning the 5' and 3' adapters and UMI-adjacent substrings of the target region with the two terminal strings of the sequence; identifying the nucleotides between the aligned target sequence and adapter sequence at each end and clustering the UMI sequences; and discarding sequences that lack UMIs at both ends and have an edit difference of less than 3 with the UMIs. In another aspect, the number of selected cluster consensus sequences is between 3 and 10; and the selected cluster size is 20 to 80.In another aspect, the method further comprises analyzing the raw nucleotide sequence data of step 1(f) or the high-accuracy consensus sequence data of step 2(l), comprising executing on a processor: receiving sequence data comprising a plurality of sequences; analyzing and merging the sample sequence data, and outputting the merged sequence; when a single-stranded or double-stranded DNA oligonucleotide donor is provided, developing a target site sequence containing a predicted result of a repair event and outputting a target prediction result; using a mapper, binning the merged sequence with the target site sequence or an optional target prediction result, and outputting a target read alignment; using an enzyme-specific position-specific scoring matrix derived from biological data based on the position of a guide sequence and a typical enzyme-specific cleavage site, re-aligning the binned target read alignment with the target site, and generating a final alignment; analyzing the final alignment within a predefined sequence distance window and identifying and quantifying mutations from the typical enzyme-specific cleavage site; outputting the final alignment, analysis and quantification result data as a table or graph. In another aspect, the purification in steps (b) and (d) comprises solid phase reversible immobilization (SPRI) purification. In another aspect, the unique molecular identifier comprises 8 to 30 nucleotides. In another aspect, the unique molecular identifier comprises 8 to 18 nucleotides. In another aspect, the universal sequence comprises 22 to 30 nucleotides. In another aspect, the barcode sequence comprises 16 to 24 nucleotides. In another aspect, the amplification in step (a) comprises at least 2 PCR cycles. In another aspect, the amplification in step (a) comprises 2 to 4 PCR cycles. In another aspect, the amplification in step (c) comprises 20 to 40 PCR cycles. In another aspect, the long read sequencing device is selected from Oxford Nanopore Technologies (ONT) MinION or PacBio Sequel II. In another aspect, the sequencing error rate is reduced by at least 15 times.

[0027] Another embodiment described herein is an analysis pipeline called CRISPAltRations. The pipeline usually accepts FASTQ files and uses FLASH to build the merged R1 / R2 consensus sequence. When processing long read sequencing data from PacBio or Oxford Nanopore Technologies platform, this initial process is not required. Instead, a target site reference is constructed, which describes the sequence of all expected target positions. Optionally, a target containing the expected result of a homology directed repair (HDR) event is constructed. Next, the merged sequence reads are aligned with the target reference sequence using minimap2, which was originally developed for rapid alignment of long reads (e.g., those generated by Oxford Nanopore Technologies MinION). Then an improved form of Needleman-Wunsch aligner (called psnw) is used to re-align the reads aligned with each target. The improved aligner allows for improved detection of insertions and deletions caused by DSB repair. All variants observed within a predefined distance of the DSB position are characterized and quantified. Finally, the results are summarized in tables and figures. The various schemes, tools and file types described are familiar and readily available to those of ordinary skill in the art. It should be understood that these schemes, tools and file types are exemplary and are not intended to be limiting. Other tools and file types may be used to practice the described processing and analysis.

[0028] In this analysis pipeline, the following improvements compared to existing methods are described. First, the use of minimap2 enables alignment of reads generated by short-read sequencers and long-read sequencers. Second, the ability to characterize perfect (i.e., correctly occurring) HDR events is improved by constructing the expected results of homology-directed repair events. Third, the use of an improved Needleman-Wunsch aligner that accepts a Cas-specific reward matrix can significantly improve insertion and deletion characterization and editing quantification percentage (%) compared to existing methods. Fourth, the graphical visualization of introduced allelic variants is improved. Fifth, the predicted repair events described in existing tools are compared with the observed repairs, and the molecular pathways involved in the repair can be described.

[0029] In one embodiment, the methods described herein have the following advantageous uses:

[0030] • Accurately characterize the indel profiles generated by DSBs.

[0031] • The percentage of editing is calculated using the fraction of reads containing indels after DSB repair. This metric (% Editing) is used to determine the effectiveness of a gRNA for CRISPR-Cas gene editing.

[0032] • Accurate characterization of the resulting indels also improves the ability to identify the percentage of cellular chromosomes in a cell population that contain frameshift mutations. Frameshift mutations modify the protein encoded by the affected gene.

[0033] • Accurately characterize the inserted sequence.

[0034] • Accurately characterize multiple mutations caused by multiple gRNA / Cas9 (i.e., ribonucleoprotein complex) delivery or dual-guide region modification.

[0035] • Analysis of indels sequenced on long-read platforms such as MinION. In addition, it allows for the phased characterization of both ends of large (>400 nt) indel events occurring after DSB repair.

[0036] • Improved visualization of results.

[0037] One embodiment described herein is a computer-implemented method for identifying and characterizing double-stranded DNA break repair sites with improved accuracy, the method comprising executing the following steps on a processor: receiving sample sequence data comprising a plurality of sequences; analyzing and merging the sample sequence data, and outputting the merged sequence; when a single-stranded or double-stranded DNA oligonucleotide donor is provided, developing a target site sequence containing a predicted result of a repair event and outputting a target prediction result; using a mapper, binning the merged sequence with the target site sequence or an optional target prediction result, and outputting a target read alignment; using an enzyme-specific position-specific scoring matrix derived from biological data applied based on the position of a guide sequence and a typical enzyme-specific cleavage site, realigning the binned target read alignment with the target site and generating a final alignment; analyzing the final alignment and identifying and quantifying mutations within a predefined sequence distance window based on the typical enzyme-specific cleavage site; and outputting the final alignment, analysis, and quantification result data as a table or graph.

[0038] In one embodiment, the edited genomic DNA is extracted and amplified using targeted multiplex PCR to enrich the on-target locus and predicted off-target locus. Amplicons are sequenced on Illumina MiSeq. When paired end reads from short read sequencing are processed using the illumina platform, the read pairs are merged into a single fragment (FLASH), mapped to the genome (minimap2), and binned according to their alignment with the expected amplicon position. The long read sequence data output of the PacBio or Oxford Nanopore Technologies platform does not require this step. After finding the cleavage site and creating a position-specific gap open / extension reward matrix, the reads in each box are re-aligned with the expected amplicon sequence to preferentially align the indel / expected indel spectrum (CRISPAltRations encoding + psnw) closer to the cleavage site for each enzyme. Annotate the indels that intersect with the window upstream or downstream of the cleavage site. The editing percentage is the sum / total number of observations of the reads containing indels.

[0039] In some embodiments, the methods described herein use minimap2, which is capable of aligning reads generated by short-read sequencers and long-read sequencers. Existing tools typically only accept short-read sequencing data, such as data generated by Illumina sequencers. Others have used long-read sequencing data to study large insertions or deletions, but it is believed that there are no independent publicly available tools. By using the minimap2 aligner, long-read data processing is partially achieved. For example, the alignment results can be visualized to show the identification of blunt molecular insertions in DNA after DSB repair.

[0040] In another embodiment, the ability to characterize a perfect HDR event is improved by constructing the expected results of the HDR event. A reference file in FASTA format contains each expected sequence target and modified sequence target. The first step in constructing the file includes creating a reference sequence index that can align the read segment with each expected structural variant. For example, if the DSB and the area targeted by the double-stranded DNA donor oligonucleotide are queried to achieve HDR, there are a variety of different possible biorepair results: complete repair, HDR-mediated repair, NHEJ repair, and NHEJ repair using repeated insertions. Other results, such as template fragments or three template inserts, are also possible. Other tools such as UDiTaS TM Use similar reference file construction methods.

[0041] In another embodiment, for short read sequences, an improved version of the Needleman-Wunsch algorithm is used to realign the read with its intended target. The method described herein improves the accuracy of alignments containing indels (as annotated in the CIGAR string of the alignment). Compared with existing methods, it significantly improves indel characterization and editing quantification%. DNA sequence aligners, such as minimap2 and Needleman-Wunsch methods, use fixed penalties for open spaces and extended spaces to measure indel alignments. The method is improved by realigning reads with their targets using position-specific gap opening and extension penalties (implemented in a tool called "psnw"), so that alignments with indels are conducive to locating them at or near the predicted DSB overlap. The position-specific matrix is ​​set to reflect the actual characterization indel spectrum of the specific Cas enzyme used for editing. Therefore, indel base alignments are most popular at or near the predicted target cleavage site (variable scoring strategy). The method can accurately realign indels, particularly those indels present in the repeat region in the reference sequence. The method improves the ability to identify the most likely outcome biologically.

[0042] In another embodiment, the method described herein collects indels near the nuclease cleavage site and marks indels that intersect the cleavage site or are within a fixed distance. Some published reports have proposed a fixed distance of 1 to 2nt, but the data supporting these choices are limited. When developing the embodiments described herein, the optimal distance (i.e., window size) around the cleavage site was studied using a set of Cas9-RNP-treated samples and paired untreated control samples. It was observed that the 4 nt window of Cas9 or the 7 nt window of Cas12a provided the highest sensitivity and provided acceptable specificity. The larger window requirement of Cas12a may be due to the mechanism of action; Cas12a achieves double-strand breaks by generating two single-strand breaks 5 bp apart (leaving "sticky" ends). Therefore, the method described herein can be extended to other nucleases (e.g., CasX) for which there is biological data to inform the target window size and enzymatic mechanism of action.

[0043] In another embodiment, the graphical visualization of allelic variation is improved. Downstream of the alignment step, some other analyses are performed that are unique to the described method. In order to generate improved visualization, the reads are replicated according to the identity of the identified insertion and deletion sequences within the CRISPR editing window after the alignment. The replicated reads are written back to the BAM file, and the frequency of each replicated read in the original read population is written to the associated BAM tag. After the file is indexed, the insertion and deletion in the replicated reads and their associated frequencies can be visualized using the commonly used IGV tool.

[0044] The utility of the systems and methods described herein was demonstrated by generating a set of synthetic 603 gRNA:amplicon pairs. At each target, 4000 read pairs (2×150 bp) were synthetically generated using simulated Illumina MiSeq v3 platform error profiles. In half of the reads, random indels were introduced based on models generated from observed editing profiles of Cas9 and Cas12a. The synthetic data were analyzed using the CRISPAltRations system described herein. By implementing the methods described herein, the ability to correctly characterize indels was improved by approximately 15% to 20%. The algorithm described herein improves accuracy because it produces a biologically informed selection of the best alignments in the target, where multiple equal-scoring alignments can be made. In addition, the methods described herein more accurately calculate the percentage of modified DNA molecules. The methods and strategies described herein are important improvements for characterizing and quantifying indels introduced after DSB repair.

[0045] Another embodiment described herein is a computer-implemented method for aligning biological sequences, the method comprising performing the following steps on a processor: receiving sample sequence data comprising a plurality of sequences; aligning the sequence data with the predicted target sequence using a matrix of enzyme-specific position-specific scores based on a specific nuclease target site; outputting the alignment results as a table or graph. In one aspect, the sequence data comprises sequences from a cell or subject population. In another aspect, the specific nuclease target sequence comprises one or more target sites of Cas9, Cas12a, or other Cas enzymes. In another aspect, the matrix uses position-specific gap opening and extension penalties.

[0046] Another embodiment described herein is a method for identifying and characterizing double-stranded DNA break repair sites with improved accuracy, the method comprising: extracting genomic DNA from a cell or tissue population of a subject; amplifying the genomic DNA using multiplex PCR to produce amplicons enriched in target site sequences; sequencing the amplicons and obtaining sample sequence data; then performing the following steps on a processor: receiving sample sequence data comprising multiple sequences; analyzing and merging the sample sequence data and outputting the merged sequence; when a single-stranded or double-stranded DNA oligonucleotide donor is provided, developing a target site sequence containing a predicted result of a repair event and outputting a target prediction result; using a mapper, binning the merged sequence with the target site sequence or an optional target prediction result and outputting a target read alignment; using an enzyme-specific position-specific scoring matrix derived from biological data based on the position of a guide sequence and a typical enzyme-specific cleavage site, re-aligning the binned target read alignment with the target site and generating a final alignment; starting from the typical enzyme-specific cleavage site, analyzing the final alignment within a predefined sequence distance window and identifying and quantifying mutations; outputting the final alignment, analysis and quantification result data as a table or graph.

[0047] Without departing from the spirit and scope of the present disclosure, the various components and methods described herein and the components or methods not shown may have many different arrangements. It should be understood that embodiments or aspects may include and be implemented by a combination of various hardware, software or electronic components. For example, various microprocessors and application specific integrated circuits ("ASICs"), as well as software in various languages, may be utilized. In addition, servers and various computing devices may be used, and may include one or more processing units, one or more computer-readable media, one or more input / output interfaces, and various connectors (e.g., a system bus) connecting the components.

[0048] It is obvious to those of ordinary skill in the relevant art that the compositions, formulations, methods, processes and applications described herein may be appropriately modified and adapted without departing from the scope of any embodiment or aspect of the present invention. The compositions and methods provided are exemplary and are not intended to limit the scope of any specified embodiment. All various embodiments, aspects and options disclosed herein may be combined in any variation or iteration. The scope of the compositions, formulations, methods and processes described herein includes all actual or potential combinations of the embodiments, aspects, options, examples and preferences described herein. The exemplary compositions and formulations described herein may omit any component, replace any component disclosed herein or include any component disclosed elsewhere herein. If the meaning of any term in any patent or publication incorporated by reference conflicts with the meaning of the term used in the present disclosure, the meaning of the term or phrase in the present disclosure shall prevail. In addition, the above discussion only discloses and describes exemplary embodiments. The specific teachings of all patents and publications cited herein are incorporated herein by reference.

[0049] Example

[0050] Example 1

[0051] Synthetic input mixture generation

[0052] To investigate the utility of UMI incorporation using a long-read sequencing platform, synthetic WT and HDR amplicons representing sequences after a hypothetical perfect HDR insertion were generated by PCR from gene templates and mixed in known ratios. This allows for known HDR frequencies in the synthetic DNA input prior to UMI sample preparation to monitor PCR bias and correct for it through UMI consensus sequence construction. To generate the input DNA mixture for UMI amplification, the synthetic amplicons were PCR amplified with a high-fidelity polymerase (Platinum SuperFi II, Thermo Fisher) and a limited number of cycles (25 amplification cycles) to reduce the possibility of polymerase errors in the input DNA mixture. The synthetic templates represent three target sites within human genes: HBB, TRAC, and SERPINC1. In addition, in the case of synthetic templates generated to represent knock-ins, the first two targets had GFP insertions of 717 bp and 729 bp in the HDR amplicons, while SERPINC1 was tested with two HDR insert lengths, 500 bp and 1971 bp (Table 1). The ratio of WT:HDR amplicons in each input mixture was quantified by fragment analyzer, qPCR was performed, sequenced using native barcodes (no PCR), using an Oxford Nanopore Technologies MinION sequencer, and analyzed using the CRISPAltRations pipeline to quantify the HDR percentage (data not shown). This represents the "expected" HDR in each sample before library preparation with UMIs incorporated.

[0053]

[0054]

[0055]

[0056]

[0057]

[0058]

[0059]

[0060]

[0061]

[0062]

[0063]

[0064]

[0065]

[0066]

[0067] Example 2

[0068] UMI library preparation and sequencing

[0069] Using input DNA (5 × 10 3 The barcoded libraries were amplified with target-specific UMI primers (Table 2) for two cycles, followed by 0.5× volume of solid phase reversible immobilization purification (SPRI; Beckman Coulter) to remove unconsumed primers. They were used as templates for the second barcoding PCR for 28 to 30 cycles, followed by 0.5× SPRI purification. Samples were visualized on a fragment analyzer, quantified by Qubit, pooled, and sequenced using a MinION sequencer from Oxford Nanopore Technologies or a PacBio Sequel II instrument to achieve a coverage depth of ≥10× / UMI (100,000 reads) / sample. Sequencing adapters were added to the final barcoded library by ligation using a kit available from the manufacturer.

[0070]

[0071]

[0072] Example 3

[0073] UMI consensus sequence construction and CRISPAltRations analysis

[0074] Our pipeline for UMI identification and consensus sequence construction was originally developed by Oxford Nanopore Technologies and is available on its github site (github.com / nanoporetech / pipeline-umi-amplicon) as (pipeline_umi_amplicon) under the Mozilla Public License Version 2.0. Several improvements have been made to the pipeline, but the general workflow remains unchanged. The key improvement is a new UMI identification method that allows processing of alternative UMI designs including 18 nt structured UMIs and random 10 nt UMIs, such as Figure 2 shown.

[0075] pipeline_umi_amplicon works by first aligning the reads to the reference genome (hg38) using minimap2, and then separating the mapped reads by target region for separate UMI identification and clustering. UMI sequences are extracted from the 5' and 3' ends of each read, and reads that do not contain both UMIs are filtered out. The UMI identification step is altered to identify UMI sequences by aligning the expected adapters and target bases surrounding the UMI to the 5' and 3' ends of each read, and then extracting bases between the alignments. The reads are clustered using vsearch, and a cluster consensus sequence is generated using medaka. The method then repeats the UMI identification, clustering, and consensus sequence construction steps on the intermediate reads to obtain higher accuracy and eliminate PCR bias that was not corrected in the first clustering and consensus sequence construction step due to the higher error rate in the UMI sequences.

[0076] The UMI pipeline generated FASTA files containing consensus sequence reads. These files were used as input to the CRISPAltRations pipeline for downstream analysis of CRISPR editing, including HDR percentage, perfect HDR percentage, and imperfect HDR percentage.

[0077] Example 4

[0078] UMI consensus sequence construction reduces the error rate

[0079] After UMI consensus construction with the required UMI cluster identity set to 80% (id80) and the minimum number of median reads per cluster group set to 10 (min10), the average error rate was calculated using an in-house built error analysis tool and decreased 16.4-fold from 8.03% to 0.491% across three sites on the Oxford Nanopore Technologies MinION. For samples sequenced on the PacBio Sequel II using circular consensus sequencing (CCS; >3 reads / molecule) with a minimum Q score of 20, the average error rate decreased 23.5-fold from 0.47% to 0.017%, indicating that incorporation of UMIs corrected for polymerase and / or sequencing errors (Table 3).

[0080]

[0081] This is also reflected in the relative fraction of total HDR reads considered “perfect HDR” by the CRISPAltRations pipeline. The high error rate of standard Oxford Nanopore sequencing typically results in <0.2% of HDR reads being called perfect, where the sequence exactly matches the expected HDR event base-by-base, even at relatively high HDR frequencies (40% to 60%). Even with the higher accuracy of PacBio HiFi sequencing, the fraction of perfect HDR reads varies between 0.15 and 0.65 and decreases as HDR inserts become longer. In general, the fraction of quantified perfect HDR reads increased after UMI consensus construction, although the extent was smallest for the longest 2 kb amplicons when sequenced using the Oxford Nanopore MinION ( Figure 3A ). However, when sequenced with HiFi reads on the PacBio Sequel II, all samples were able to report >80% of total HDR reads as perfect ( Figure 3B ). For both the Oxford Nanopore Technologies platform and the PacBio platform, the UMI min10 parameter (meaning that 10 reads were required to form a cluster) showed the strongest improvement over the raw reads compared to the min3 parameter. Additional UMI parameters were investigated, including a minimum UMI sequence identity within a cluster of 70% (id70) or 90% (id90), but did not further improve performance (data not shown). Further investigation of the minimum number of intermediate reads is planned.

[0082] Example 5

[0083] UMI consensus sequence construction for correcting PCR bias

[0084] PCR bias became more pronounced as the HDR insert size increased relative to the WT amplicon length. The longest HDR insert in this test set was 1971 bp at the SERPINC1 locus, which is nearly equal to the WT amplicon length of 1991 bp. After library preparation and sequencing using the Oxford Nanopore Technologies platform for this sample, the raw HDR rate dropped from the expected 31.1% HDR to 19.2%. After UMI consensus construction, the total HDR percentage increased to 23.2% (min10) and 26.6% (min3), closer to the expected HDR frequency for this sample. Further investigation of coverage depth requirements and ideal UMI consensus construction parameters to identify optimal sequencing and analysis conditions may improve the robustness of this method in terms of error correction and PCR bias correction.

Claims

1. A method for improving the accuracy of long-read sequencing, the method comprising: Generate sequencing libraries, including: (a) amplifying the locus using primers containing a unique molecular identifier and a universal sequence to generate an initial product; (b) purifying the initial product; (c) amplifying the initial product using primers comprising a sequence complementary to the universal sequence and a barcode sequence to generate a barcoded product; (d) purifying the barcoded product to produce a purified barcoded product; (e) pooling the purified barcoded products to produce pooled barcoded products; and (f) Sequencing the pooled barcoded products using a long-read sequencing device to generate raw nucleotide sequence data.

2. The method according to claim 1, further comprising executing on a processor: (g) receiving raw nucleotide sequence data; (h) aligning the raw nucleotide sequence data with a reference amplicon to generate a mapped sequence; (i) identifying and separating mapped sequences by target region to generate multiple sets of target region sequences; (j) For each set of target region sequences: (i) analyzing the target region sequences for unique molecular identifiers and discarding the target region sequences lacking unique molecular identifiers; (ii) clustering the target region sequences containing unique molecular identifiers to generate clustered target region sequences and clustered consensus sequences; (iii) analyzing and filtering the cluster target region sequences, discarding sequences with a cluster consensus number less than a selected number, and downsampling clusters larger than a selected cluster size to the selected cluster size; (iv) generating an initial target sequence consensus sequence; (k) repeating step (j) for the initial target sequence consensus sequence to generate a high-accuracy consensus sequence for each clustering group, and correcting amplification bias by clustering groups that were not similar enough to be clustered in the first round; (l) Output high-accuracy consensus sequence data.

3. The method of claim 2, wherein step (j)(i) comprises: Aligning the 5' and 3' adapters and UMI-adjacent substrings of the target region with the two terminal strings of the sequence; Identifying nucleotides between the aligned target sequence and adapter sequence on each end and clustering UMI sequences; and Sequences lacking a UMI at both ends and with an edit difference of less than 3 from the UMI were discarded.

4. The method of claim 2, wherein the number of selected cluster consensus sequences is between 3 and 10; and the selected cluster size is between 20 and 80.

5. The method according to any one of claims 1 to 4, further comprising analyzing the raw nucleotide sequence data from claim 1(f) or the high-accuracy consensus sequence data from claim 2(l), the method comprising executing on a processor: receiving sequence data comprising a plurality of sequences; Analyze and merge sample sequence data, and output merged sequences; When a single-stranded or double-stranded DNA oligonucleotide donor is provided, a target site sequence containing a predicted outcome of a repair event is developed and a target prediction outcome is output; Using a mapper, bin the merged sequences using the target site sequence or optional target prediction results and output the target read alignment; realigning the binned target read alignments to the target sites using an enzyme-specific position-specific scoring matrix derived from biological data applied based on the guide sequences and the locations of typical enzyme-specific cleavage sites and generating a final alignment; Starting from typical enzyme-specific cleavage sites, the final alignment is analyzed and mutations are identified and quantified within predefined sequence distance windows; Output the final comparison, analysis and quantification result data as tables or graphs.

6. The method of claim 1, wherein the purification in step (b) and step (d) comprises solid phase reversible immobilization (SPRI) purification.

7. The method of claim 1, wherein the unique molecular identifier comprises 8 to 30 nucleotides.

8. The method of claim 1, wherein the unique molecular identifier comprises 8 to 18 nucleotides.

9. The method of claim 1, wherein the universal sequence comprises 22 to 30 nucleotides.

10. The method of claim 1, wherein the barcode sequence comprises 16 to 24 nucleotides.

11. The method of claim 1, wherein the amplification in step (a) comprises at least 2 PCR cycles.

12. The method of claim 1, wherein the amplification in step (a) comprises 2 to 4 PCR cycles.

13. The method of claim 1, wherein the amplification in step (c) comprises 20 to 40 PCR cycles.

14. The method of claim 1, wherein the long-read sequencing device is selected from Oxford Nanopore Technologies (ONT) MinION or PacBio Sequel II.

15. The method of claim 1, wherein the sequencing error rate is reduced by at least 15-fold.

Images