Methods and systems for analyte-ligand interaction analysis

Abstract

In various aspects, the invention relates to methods 700, 800, and systems 900 for determining one or more interaction parameters associated with an interaction between an analyte and a ligand. The method comprises contacting 704, 804; 705, 806) a sensor surface having a ligand immobilized thereon with one or more samples comprising an analyte; recording 706, 806 a sensor response indicative of binding of the analyte to the binding site of the ligand; and fitting 716, 808 the recorded sensor responses to an interaction model in order to determine one or more interaction parameters associated with an interaction between the analyte and the ligand, where the fitting is constrained by predetermined upper and lower limits applied to an analyte saturation parameter associated with the interaction model.

Description

Technical Field

[0001] The present disclosure relates to the analysis of interactions between analytes and ligands on sensor surfaces and, more particularly, to systems and methods that enable improved determination of interaction parameters associated with such interactions. Background Art

[0002] Analytical sensor systems capable of monitoring interactions between molecules (such as biomolecules) in real time are attracting increasing attention. Such systems are generally capable of determining one or more of the binding, kinetics, affinity, specificity, and concentration of a molecule ("analyte") contained in a sample solution. Optical biosensors are particularly useful for this purpose and are often referred to as interaction analysis sensors or biospecific interaction analysis sensors. Representative biosensor systems of this type are sold by Cytiva. The instrument uses surface plasmon resonance (SPR) to detect the interaction between molecules on a sensing surface without the need for any labels. As the corresponding sample passes over the sensor surface, the progress of the binding can be measured, providing a direct reflection of the rate at which the intermolecular interaction occurs.

[0003] From such as The typical output of a system like this is a graph or curve that describes the progression of a molecular interaction over time, including parts of the association phase and parts of the dissociation phase. Such a binding curve, usually displayed on a computer screen, is often called a "sensorgram". The system (and similar sensor systems) can not only determine in real time the presence and concentration of a specific analyte in a sample without the use of labels (and usually without purification of the substances involved), but can also determine in real time additional interaction parameters, including the kinetic rate constants for binding (association) and dissociation in the molecular interaction, as well as the affinity of the interaction being evaluated. The association rate constant (k) can be obtained by fitting the obtained kinetic data for one or preferably several different sample analyte concentrations to a mathematical description of the interaction model in the form of a differential equation. a ) and the dissociation rate constant (k d ). Affinity (expressed as the association equilibrium constant K A or dissociation equilibrium constant K D ) can be calculated from the association and dissociation rate constants.

[0004] However, due to weak binding, it can often be difficult to obtain definitive kinetic data, and therefore it is often more reliable to measure the affinity of the analyte by equilibrium binding analysis, which involves determining the level of binding for a range of analyte concentrations at equilibrium or steady state, which is believed to be achieved at or near the end of the association phase of the binding interaction.

[0005] However, a problem arises in that determining affinity in this manner by equilibrium binding analysis has traditionally relied on being able to obtain sensor response data across a wide range of analyte concentrations. Specifically, conventional analytical techniques require that the dissociation equilibrium constant, K, be approached. D At concentrations close to K, a sensor response value for the analyte can be obtained. However, in practice, such data may be difficult to obtain due to practical limitations, such as the fact that above a given concentration the analyte may begin to precipitate or aggregate, making it impossible to read the sensor response accurately. Therefore, for sensors close to K D It may not always be possible to obtain an undisturbed sensor response for a wide range of analyte concentrations. Instead, responses may only be available for a limited range of analyte concentrations. As an example, this situation may often occur during the screening of drug candidates. In this case, the parameters used to calculate the interaction parameters (e.g., K D , K A ) are insufficient and produce unrealistic and imprecise estimates of these values. Although some progress has been made in providing alternative solutions, these solutions still cannot provide completely accurate data, as will be described in further detail below.

[0006] Thus, it can be seen that existing processes for determining interaction parameters of analyte-ligand interactions suffer from significant shortcomings. It would be advantageous to provide systems and methods that address one or more of these problems, either individually or in combination. Summary of the Invention

[0007] This Summary introduces concepts that are described in greater detail in the Detailed Description. It should not be used to identify essential features of the claimed subject matter, nor should it be used to limit the scope of the claimed subject matter.

[0008] According to one aspect of the present disclosure, a method for determining one or more interaction parameters associated with an interaction between an analyte and a ligand is provided. The method comprises contacting a sensor surface having a ligand immobilized thereon with one or more samples containing the analyte, and recording a sensor response indicative of binding of the analyte to a binding site of the ligand.

[0009] Preferably, the sensor surface is contacted with three or more samples containing different concentrations of the analyte. Injecting a series of samples of different analyte concentrations in this way can be called a concentration series and provides more reliable results than using a single analyte concentration.

[0010] Preferably, to improve the accuracy of the results, the responses recorded are the response levels for each analyte sample recorded at equilibrium (or as close to equilibrium as possible), which in practice is usually assumed to be at the end of the correlation phase.

[0011] The method further comprises fitting the recorded sensor responses to an interaction model to determine one or more interaction parameters associated with the interaction between the analyte and the ligand. Advantageously, the fitting is subject to predetermined upper and lower limits applied to the analyte saturation parameters associated with the interaction model.

[0012] Fitting the recorded sensor responses to an interaction model is a known method for determining interaction parameters. Surprisingly, however, the inventors have recognized that the accuracy of this method may be significantly improved by constraining the fitting in the manner described above, and this will be explained in further detail herein. Specifically, by applying upper and lower limits to the analyte saturation parameter associated with the interaction model, the fitting process is constrained to produce values ​​within certain boundaries. The inventors believe that this constraint produces a more accurate estimate of the interaction parameter than any existing method. In particular, the inventors have recognized that the disclosed method effectively provides a middle ground between previous methods, which either rely too much on recorded response data or, at the other extreme, use a fixed estimate of the analyte saturation parameter and thereby give too little weight to the actual recorded response data. The currently disclosed method represents a balance between these competing methods, and surprisingly, the inventors believe that this method provides a more accurate determination of affinity.

[0013] The one or more interaction parameters determined using the disclosed methods may include one or more of the following: the association equilibrium constant K of the interaction A ; and the dissociation equilibrium constant K of the interaction D These interaction parameters provide valuable insights into the binding behavior of an analyte in the presence of a ligand, particularly regarding the affinity of the analyte for the ligand in question. This data can be used to inform a variety of important assays and analyses, such as drug candidate screening, antibody profiling, and quality control.

[0014] As described above, the fitting of the sensor response data to the interaction model is subject to predetermined upper and lower limits applied to the analyte saturation parameter associated with the interaction model. In one example, the analyte saturation parameter to which the upper and lower limits are applied may be the analyte saturation response value R associated with the interaction model used in the fitting process. max . R max represents the maximum possible sensor response for a given concentration of analyte, i.e. the response that would be expected if the analyte had bound to all available binding sites of the ligand on the sensor surface. As will be described in further detail below, the present inventors have recognized that, surprisingly, by applying R max The upper and lower limits to constrain the fit are considered in determining factors such as K D and KA provides increased precision in interaction parameters such as .

[0015] Fitting sensor response data to an R-based max An interaction model based on the target occupancy To is not the only option. An alternative is to use an interaction model based on the target occupancy To. In this case, the analyte saturation parameter to which the upper and lower limits apply is the target occupancy value To at saturation. sat The target occupancy can be thought of as expressing the analyte response as a percentage (or fraction) of the response that would be expected at saturation. Thus, in a sensor with a response of R max When the saturation of sat =100% (or a fraction of 1) because the response will be maximal when there is maximum analyte-ligand binding (i.e., at saturation). When fitting the sensor response data to an interaction model based on target occupancy, the inventors again recognized that using the target occupancy To applied to saturation sat The upper and lower limits of K D and K A Better prediction of interaction parameters such as

[0016] In one example, the upper and lower limits applied to the analyte saturation parameter can be set by the user based on their knowledge of the particular assay. Alternatively, however, the disclosed method can include determining an estimate of the analyte saturation parameter and setting at least one of the upper and lower limits based on the estimate of the analyte saturation parameter. In other words, an initial estimate of the analyte saturation parameter can be determined, and this estimate can then be used as a starting point around which to set the upper and lower limits. This approach means that the fitting process is anchored to a best guess estimate of the analyte saturation parameter, which has been previously estimated based on theoretical or experimental data. This approach increases the likelihood of accurately determining the interaction parameter compared to simply relying on the user to provide an estimate based on their knowledge of the analyte and its possible interaction with the ligand.

[0017] In one example, at least one of the upper and lower limits is set as a percentage of the estimated value of the analyte saturation parameter. In other words, if the estimated value represents 100%, the fit can be constrained between the upper and lower limits, which are defined as percentages of this initial estimate. The inventors have recognized that a particularly suitable lower limit is 75% of the estimated analyte saturation parameter. It has been found that this constraint provides a highly accurate estimate of the interaction parameter. Similarly, the inventors have recognized that a particularly suitable upper limit is 150% of the estimated analyte saturation parameter. It has also been found that this constraint provides a highly accurate estimate of the interaction parameter. Therefore, in a particularly advantageous arrangement, the fitting process is restricted to using values ​​between 75% and 150% of the previously determined estimate of the analyte saturation parameter. This constraint prevents the fitting process from considering values ​​outside this interval of the analyte saturation parameter, thereby constraining the fitting process within predefined boundaries. As described above, it has been found that this approach improves the accuracy of interaction parameter determination. Although it has been found that the limits of 75% and 150% are particularly advantageous, when the lower limit is between 75% and 99% of the estimated value, and even when the lower limit is between 25% and 99% of the estimated value, still the benefit in accuracy can be provided. Similarly, when the upper limit is between 101% and 150% of the estimated value, and even when the upper limit is between 101% and 250% of the estimated value, a benefit in accuracy is provided. Therefore, the lower limit can be set to between 75% and 99% of the estimated value, or alternatively, be set to between 25% and 99% of the estimated value. The upper limit can be set to between 101% and 150% of the estimated value, or alternatively be set to between 101% and 250% of the estimated value. All of these percentage ranges are considered to provide benefits in determining accurate interaction parameters.

[0018] As already mentioned, if the upper and lower limits used to constrain the fitting process are based on a predetermined estimate of the analyte saturation parameter, then the predetermined estimate can be determined in a variety of ways. In one example, the estimate of the analyte saturation parameter is theoretically determined based on known models and equations regarding the relationship between the analyte and the ligand. In this case, determining the estimate of the analyte saturation parameter can include calculating a theoretical estimate of the value based on:

[0019] Molecular weight of the ligand Mw lig ;

[0020] Molecular weight Mw of the analyte ana ;and

[0021] Ligand response R lig .

[0022] Ligand response R ligIt can also be referred to as the immobilization level of the ligand. The advantage of this theoretical approach is that an initial estimate of the saturation parameter of the analyte can be determined theoretically without the need to control the analyte or perform any experiments, which saves time and cost.

[0023] In one example, if the interaction model used in the fitting process is based on R max The predetermined estimate of the analyte saturation parameter can be R max The estimated theoretical value of R max This theoretical estimate of can be determined using the following equation:

[0024]

[0025] Alternatively, if the interaction model used in the fitting process is based on target occupancy (To), the predetermined estimate of the analyte saturation parameter (To sat ) can be defined as 1. The sample response (R A ) can be calculated based on the target occupancy (To A )express:

[0026]

[0027] While theoretically determining an estimate of the analyte saturation parameter is one option, another option is to experimentally determine the value using a control analyte that is assumed to interact with the ligand in a manner similar to how the analyte of interest interacts. In this case, determining an estimate of the analyte saturation parameter can include contacting the sensor surface with the control analyte, recording a sensor response indicative of binding of the control analyte to the ligand binding site, and determining a control analyte saturation parameter for the control analyte. For example, R maxC Control analytes may be determined.

[0028] Once the control analyte saturation parameter has been determined, it can be converted to the corresponding saturation parameter for the analyte of interest. For example, R maxC (for control analytes) can be converted to R maxA (for the analyte of interest.) Methods for performing this conversion are described in more detail below.

[0029] As already mentioned, the method can include receiving user input that sets upper and lower limits. In other words, the user can set upper and lower limits around the analyte saturation parameter, by which the fitting process is constrained. As described above, in one example, the user can simply set the limits based on existing knowledge of the interaction being analyzed. Alternatively, the system can store a predetermined estimate of the analyte saturation parameter (e.g., an estimate determined in one of the theoretical or experimental ways described above). In this case, the predetermined estimate can serve as a default value, and the user input can be used to set the upper and lower limits relative to the predetermined estimate.

[0030] In one example, the user can enter values ​​for upper and lower limits, such as in units of response (RU) or target occupancy. For example, if the predetermined estimate is an R of 14 RU max If the user input is set to +2RU relative to the estimated value, the upper limit can be set to +2RU relative to the estimated value, and the lower limit can be set to -2RU relative to the estimated value. In this case, the fitting process will be constrained to result in R values ​​between 12 and 16RU. max In another example, the user can enter a percentage value that determines an upper limit and a lower limit relative to a predetermined estimate of the analyte saturation parameter. For example, the user can set the upper limit at 150% of the estimated value of the analyte saturation parameter and the lower limit at 75% of the estimated value of the analyte saturation parameter. In the aforementioned R max In the example of a "default" value of 14 RU, these percentage limits would therefore restrict the fitting process to fits that result in values ​​between 10.5 and 22.5 RU. Another practical approach is to set the lower limit to the response of the sample with the highest analyte concentration.

[0031] Once the upper and lower bounds for the fitting process have been set, it will be appreciated that a variety of fitting algorithms can be used to fit the sensor response data to the interaction model. The fitting process typically involves using an algorithm to iteratively "guess" R max and K D The values ​​of and are used to draw response curves using these guessed values ​​and the interaction model, the details of which are outlined below. It is then possible to determine how well each set of guessed values ​​fits the actual recorded response data. max The values ​​are constrained by the aforementioned upper and lower limits. In one example, fitting includes applying a minimum chi-square estimate to the recorded sensor responses to determine the quality or closeness of the fit to the response data, ie, determining how well each iteration of guess values ​​fits the response data.

[0032] In one example, the predetermined interaction model used in the fitting process includes the expression:

[0033]

[0034] in

[0035] R eq is the sensor response recorded at equilibrium,

[0036] C is the analyte concentration,

[0037] R max is the analyte saturation parameter, and

[0038] K D is the dissociation equilibrium constant of the interaction.

[0039] Thus, in this example, fitting the sensor response data to the interaction model involves algorithmically fitting the sensor response data (which provides R for corresponding values ​​of C eq The value of R max and K D As mentioned above, this involves the algorithm iteratively changing R max and K D The two are combined to find the combination that gives the best fit, as measured by a fit parameter such as chi-square. As mentioned above, the fit is constrained in that the fit does not involve R values ​​outside of the defined upper and lower limits. max Once the fitting is complete, i.e. once the best fit is found within the constraints of the upper and lower limits and the interaction model, the interaction parameter K has been determined. D The best estimate of K A It can also be calculated.

[0040] The sensor system used to perform the disclosed method may be a surface plasmon resonance (SPR) system. In this case, recording a sensor response indicative of binding of an analyte to a binding site of a ligand may be based on surface plasmon resonance.

[0041] The sensor system used to perform the disclosed method may be an evanescent wave sensing system. In this case, recording a sensor response indicative of binding of an analyte to a binding site of a ligand may be based on evanescent wave sensing.

[0042] According to another aspect of the present disclosure, a sample analysis system for determining an interaction parameter is disclosed. The system includes a sensing surface configured to detect a binding interaction between an analyte and a ligand at the sensing surface; and a computing device configured to execute any of the methods disclosed herein.

[0043] According to further aspects of the present disclosure, a computer is disclosed, wherein the computer is configured to cause a sample analysis system to perform any of the methods disclosed herein.

[0044] According to yet another aspect of the present disclosure, a computer-readable storage medium is disclosed, wherein the computer-readable storage medium includes instructions that, when executed by a computer, cause the computer to cause a sample analysis system to perform any of the methods disclosed herein. BRIEF DESCRIPTION OF THE DRAWINGS

[0045] Illustrative implementations of the present disclosure will now be described, by way of example only, with reference to the accompanying drawings. In the drawings:

[0046] Figure 1 shows a schematic illustration of an SPR-based biosensor system;

[0047] Figure 2 A representative sensorgram showing the detector response versus time for interactions between molecules occurring at the sensor surface is shown;

[0048] Figure 3 The relationship between the equilibrium binding response and concentration of the analyte over a wide concentration range is shown;

[0049] Figure 4 It shows that by Figure 3 The curve generated by fitting the data points to the interaction model;

[0050] Figure 5 The equilibrium binding response versus concentration of the analyte over a narrow concentration range is shown;

[0051] Figure 6 It shows that by Figure 5 The curve generated by fitting the data points to the interaction model;

[0052] Figure 7 shows the saturation response R based on the control analyte maxC The analyte saturation response R maxA Methods;

[0053] Figure 8 A method for determining one or more interaction parameters associated with an interaction between an analyte and a ligand according to the present disclosure is shown;

[0054] Figure 9 shows a schematic illustration of a Biacore T200 from Cytiva, which is an example of a sample analysis system that can be used to implement the disclosed methods; and

[0055] Figure 10 Components of an example computer device that may be used to implement the methods described herein are shown.

[0056] Like reference numerals denote like features throughout the specification and drawings. DETAILED DESCRIPTION

[0057] refer to Figure 1 and Figure 2 This detailed description describes the basic principles of biosensors (also called sample analysis systems) and their methods of operation. Figures 3 to 7 The shortcomings of existing methods for determining interaction parameters using such systems are described. Figure 8 An improved method for determining interaction parameters is described. Finally, Ref. Figure 9 and Figure 10 , which describes components of an example sample analysis system and an example computer device that can be used to implement the methods disclosed herein.

[0058] The methods disclosed herein generally involve determining one or more interaction parameters associated with the interaction between an analyte and a ligand, such as the correlation (K A ) or dissociation (K D ) equilibrium constant. As mentioned above, conventional methods for determining this parameter present problems because they are only reliable when data are available for a wide range of analyte concentrations. More specifically, such methods rely on having a finite element that covers (preferably exceeds) the analyte dissociation equilibrium constant K D In situations where only data for a narrow range of analyte concentrations are available (as might occur in various situations where high concentrations of analytes cannot be obtained or reliably studied), these existing methods fall short because the fitting process leads to unrealistic and imprecise estimates of the interaction parameters.

[0059] One proposed solution to this drawback is to utilize a control analyte. This approach is described in European Patent No. 2507618 to GE Healthcare Bio-Sciences AB, the contents of which are incorporated herein by reference in their entirety. In this approach, a steady-state binding assay is first performed on the control analyte to obtain a saturation response for the control analyte, denoted as R. maxC This control parameter can then be used to estimate the corresponding parameter for the analyte of interest (denoted as R maxA ). Once R maxA Once estimated, this value is used as a fixed value in the interaction model to enable determination of the interaction parameter of interest (e.g. K D , K A ).

[0060] This approach offers an improvement over previous methods that simply relied on free fitting of response data to an interaction model without further constraints or restrictions on the analyte saturation parameters. However, the improved method of EP2507618 is not without its own problems. First, this approach requires finding a control analyte, which is not always simple. The control analyte needs to be similar to the analyte of interest in its binding properties in order to be able to obtain the desired results from R. maxC to R maxA Finding such suitable control analytes can require extensive experimental effort, as suitable control analyte candidates must be identified and tested. Often, control analyte candidates perform poorly and are less effective than theoretically predicted, meaning that a large number of candidates must often be tested. In practice, even in complex laboratory settings, finding suitable control analytes can take weeks or even months. This presents a significant barrier to conducting large numbers of analyses, as many must be preceded by research to identify suitable control analytes. In some cases, no suitable control analytes are available.

[0061] Even in cases where a suitable control analyte can be found, the present inventors have recognized that R maxC to R maxA The conversion of is not always particularly successful because it relies on a number of assumptions about the similarity of the controlling analyte to the analyte of interest. In practice, these assumptions may not hold as they do in theory, which means that when this approach is used, the interaction parameters subsequently determined (e.g., K D , K A ) is not always particularly accurate.

[0062] The presently disclosed invention attempts to address the above-mentioned shortcomings. In particular, the inventors have recognized that there is another possibility for determining the interaction parameter of the interaction. The inventors believe that, surprisingly, if the fit of the response data to the interaction model is affected by the application of the analyte saturation parameter associated with the interaction model (e.g., R max ) interval (i.e., upper and lower bounds), a more accurate estimate of the interaction parameter can be obtained.

[0063] The present inventors have determined that the present method effectively serves as an intermediate position between the traditional method (freely fitting data to the interaction model without constraints on the analyte saturation parameter) and the method of EP2507618, which constrains fitting with a fixed analyte saturation parameter determined using a control analyte. Advantageously, the disclosed method is believed to retain the benefits of the method of EP2507618, i.e., it is able to provide an accurate estimate of the interaction parameter even when response data are only available for a limited range of analyte concentrations. However, importantly, the disclosed method does not rely on the use of control parameters compared to the EP2507618 method. Most importantly, by achieving a balance between existing methods and performing a fitting process constrained by intervals (i.e., upper and lower limits) rather than fixed values, the present inventors believe that more accurate estimates of the interaction parameters can be obtained than would be possible using any of the previously described methods.

[0064] To aid understanding, each of the above methods will now be described in more detail. First, the principles of the underlying sensor technology used for interaction analysis will be described. Next, a conventional free-fitting approach to fitting response data to an interaction model will be outlined, which places no restrictions on the analyte saturation parameter. Then, an alternative approach of EP2507618 will be explained, which uses a fixed value R for the saturation response. max Finally, the novel method of the present disclosure will be summarized and compared with existing methods, which improves both previous methods.

[0065] Background – Basic Sensor Technology

[0066] As described above, the present disclosure relates to the evaluation of binding response data obtained for an analyte at multiple concentrations, typically at equilibrium (steady state). Thus, one or more interaction parameters of the interaction can be determined. Typically, experimental binding data are obtained using sensor-based techniques that study molecular interactions and present results in real time as the interaction progresses. To aid understanding, some brief background on such sensor-based techniques will now be provided.

[0067] Chemical sensors or biosensors (also referred to herein as "sample analysis systems") are typically based on label-free techniques that detect changes in properties of the sensor surface, such as mass, refractive index, or thickness of a fixed layer, although some sensors also rely on some type of label. Typical sensor detection technologies include, but are not limited to, mass detection methods such as optical, thermo-optical, and piezoelectric or acoustic wave methods (including, for example, surface acoustic wave (SAW) and quartz crystal microbalance (QCM) methods), as well as electrochemical methods such as potentiometric, conductometric, amperometric, and capacitive / impedance methods. With respect to optical detection methods, representative methods include those that detect mass surface concentrations (e.g., reflectometry methods), including both external and internal reflection methods that are angle-, wavelength-, polarization-, or phase-resolved, such as evanescent wave ellipsometry and evanescent wave spectroscopy (EWS or internal reflection spectroscopy), both of which can include evanescent field enhancement by surface plasmon resonance (SPR), Brewster angle refractometry, critical angle refractometry, frustrated total reflection (FTR), scattered total internal reflection (STIR) (which can include scattering-enhancing labels), optical waveguide sensors; external reflection imaging, evanescent wave-based imaging (e.g., critical angle-resolved imaging), Brewster angle-resolved imaging, SPR angle-resolved imaging, etc. In addition, photometry and imaging / microscopy methods based on, for example, surface-enhanced Raman spectroscopy (SERS), surface-enhanced resonant Raman spectroscopy (SERRS), evanescent wave fluorescence (TIRF), and phosphorescence, "by themselves" or in combination with reflectometry methods, as well as waveguide interferometry, waveguide leaky mode spectroscopy, reflection interferometry (RIfS), transmission interferometry, holographic spectroscopy, and atomic force microscopy (AFR) can be mentioned.

[0068] Commercially available biosensors include the The system instrument, manufactured and sold by Cytiva of Uppsala, Sweden, is based on surface plasmon resonance (SPR) and allows real-time monitoring of surface binding interactions between a bound ligand and an analyte of interest. In this context, a "ligand" is a molecule with a known or unknown affinity for a given analyte and includes any capture agent or trapping agent immobilized on the sensor surface, while an "analyte" includes any specific binding partner of the ligand that is typically injected and flows over or past the sensor surface.

[0069] Although in the following detailed description, the present invention is in the context of SPR spectroscopy, more specifically in While this invention is described in the context of a system, it should be understood that the invention is not limited to this detection method. Rather, any affinity-based detection method may be used in which an analyte binds to a ligand immobilized on a sensing surface, and a change in the sensing surface can be measured that quantitatively indicates binding of the analyte to the immobilized ligand.

[0070] The SPR phenomenon is well known and suffices to say that SPR occurs when light is reflected at an interface between two media of different refractive indices under certain conditions, and this interface is covered by a metal film (usually silver or gold). In the instrument, the medium is the sample and the glass of the sensor chip, which is in contact with the sample through a microfluidic flow system. The metal film is a thin layer of gold on the surface of the chip. SPR causes the intensity of the reflected light to decrease at a specific reflection angle. The angle of minimum reflected light intensity varies with the refractive index of the surface opposite to the side of the reflected light. The sample side of the system.

[0071] The schematic diagram of the system is as follows Figure 1 As shown in Figure 1, a sensor chip 101 comprises a gold film 102 supporting a capture molecule (ligand) 103 (e.g., an antibody). The capture molecule 103 is exposed to a sample flow containing an analyte 104 (e.g., an antigen) via a flow channel 105. Monochromatic p-polarized light 106 from a light source 107 (e.g., an LED) is coupled by a prism 108 to a glass / metal interface 109, where it is totally reflected. The intensity of the reflected light beam 110 is detected by an optical detection unit 111 (e.g., a photodetector array).

[0072] A reference to the invention can be found in U.S. Patent No. 5,313,264. Technical aspects of the instrument and a detailed discussion of the SPR phenomenon. More detailed information on matrix coatings for biosensor sensing surfaces is given in, for example, U.S. Patent Nos. 5,242,828 and 5,436,161. A detailed discussion of technical aspects of biosensor chips used in conjunction with the instrument can be found in US Patent No. 5,492,840. The above-referenced publications, as well as any other publications, patent applications, patents, or other references mentioned in this disclosure, are incorporated herein by reference in their entirety.

[0073] When molecules in the sample (analytes) bind to capture molecules (ligands) on the surface of the sensor chip, the concentration of the surface changes, and therefore the refractive index also changes, and an SPR response is detected. During the interaction process, plotting the response versus time will provide a quantitative measure of the interaction process. This graph, or kinetic or binding curve (binding isotherm), is often called a sensorgram, and is sometimes also called an "affinity trace" or "affinity map" in the art. In this system, the SPR response is expressed in resonance units (RU). One RU represents a 0.0001° change in the angle of minimum reflected light intensity, which for most proteins and other biomolecules corresponds to approximately 1 pg / mm on the sensor surface. 2When a sample containing an analyte contacts the sensor surface, the capture molecules (ligands) bound to the sensor surface interact with the analyte in a step called "association." This step is indicated on the sensorgram by an increase in response (RU) when the sample initially contacts the sensor surface. Conversely, "dissociation" typically occurs when the sample flow is replaced by, for example, a buffer flow. This step is typically indicated on the sensorgram by a decrease in response (RU) over time as the analyte dissociates from the surface-bound ligand.

[0074] Representative sensorgrams (binding curves) of reversible interactions on the sensor chip surface are shown in Figure 2 is presented in . The sensorgram represents the interaction involving an immobilized capture molecule (ligand), such as an antibody, interacting with a binding partner (analyte) in the sample. Binding curves generated by biosensor systems based on the other detection principles mentioned above will have a similar appearance. The vertical axis (y-axis) indicates the response (here in resonance units RU), and the horizontal axis (x-axis) indicates the time (here in seconds). Initially, a buffer is passed over the sensing surface, giving a baseline response K in the sensorgram. During sample injection, an increase in signal is observed due to binding of the analyte to the ligand. This part L of the binding curve is often referred to as the "association phase". Ultimately, a steady-state condition is reached at the end or near the end of the association phase, when the resonance signal stabilizes at M (however, this state is not always achieved). It should be noted that the term "steady state" is used synonymously with the term "equilibrium" in this article (in other contexts, the term "equilibrium" can be reserved to describe an ideal interaction model, because in practice, binding can remain unchanged over time even if the system is not in equilibrium). At the end of the sample injection, the sample is replaced by a continuously flowing buffer, and the decrease in the signal reflects the dissociation or release of the analyte from the surface. This portion N of the binding curve is generally referred to as the "dissociation phase". The analysis is optionally terminated by a regeneration step, wherein if the length of time to complete dissociation in the buffer becomes impractical, a solution capable of removing the bound analyte from the surface while (ideally) maintaining ligand activity is injected onto the sensor surface. This is indicated in the O portion of the sensorgram. The injection of buffer restores the baseline K, and the surface is now ready for a new analysis.

[0075] From the curves for the association and dissociation phases, L and N, respectively, information about the binding and dissociation kinetics is obtained. The height of the resonance signal at M represents the affinity (the response resulting from the interaction associated with changes in surface mass concentration). This will now be explained in more detail below in the context of the various aforementioned methods for determining the interaction parameters of analyte-ligand interactions on the sensor surface.

[0076] Interaction parameter determination – Background

[0077] To aid understanding, some background derivation of the interaction model that underpins the disclosed method will first be set forth.

[0078] First, we assume a reversible reaction between analyte A and surface-bound (immobilized) capture molecules or ligands B that is not limited by diffusion or mass transfer and obeys pseudo-first-order kinetics:

[0079]

[0080] This interaction model (often referred to as the Langmuir model) assumes that the analyte (A) is both monovalent and homogeneous, the ligand (B) is homogeneous, and all binding events are independent. It has been determined that these assumptions actually hold true for the vast majority of cases, making this a valid assumption in both the free fit method and the method of the present disclosure.

[0081] During analyte injection, the rate of change of the surface concentration of analyte A (which is equal to the rate of change of concentration of the formed complex AB) is the sum of the rates of association and dissociation of analyte A:

[0082]

[0083] where [A] is the concentration of analyte A, [B] is the concentration of ligand B, [AB] is the concentration of reaction complex AB, and k a is the association rate constant, and k d is the dissociation rate constant.

[0084] After time t, the concentration of unbound ligand B on the surface is [BT]-[AB], where [BT] is the total or maximum concentration of ligand B. Substituting into equation (1) gives:

[0085]

[0086] In terms of detector response units (AB detected), this can be expressed as:

[0087]

[0088] where R is the response at time t in resonance units (RU), C is the initial or bulk concentration of free analyte (A) in solution, and R max is the response (in RU) that would be obtained if the analyte (A) had bound to all the ligands (B) on the surface, also known as the saturation response.

[0089] Rearranging equation (3) gives:

[0090]

[0091] Where R is the response in resonant units (RU). In integral form, the equation is:

[0092]

[0093] Now, according to equation (4), if dR / dt is plotted against the bound analyte concentration R, the slope is –(k a C+k d ), and the vertical intercept is k a CR max If the bulk concentration C is known and the saturation response R max Once ρ is determined (e.g. by saturating the surface with a large excess of analyte), the association rate constant k can be calculated. a and the dissociation rate constant k d .

[0094] However, a more convenient method is to fit the integral function (5), or to numerically calculate and fit the differential equation (4), preferably with the aid of a computer program. This provides an alternative method for determining R when saturating the surface with a large excess of analyte is not feasible. max alternative method.

[0095] k d It can also be determined in the following way. The dissociation rate can be expressed as:

[0096]

[0097] And in integrated form:

[0098]

[0099] where R0 is the response at the beginning of the dissociation phase (when buffer wash of the surface begins).

[0100] Equation (6) can be linearized:

[0101]

[0102] And the curve of ln[R / R0] versus t will have a slope of -k d However, it is more convenient to determine the dissociation rate constant k by fitting the exponential rate equation (7) d In order to obtain reliable kinetic constants, the above analysis is typically repeated for a number of different analyte concentrations and, where appropriate, also for at least one other ligand density on the sensor surface.

[0103] The affinity is determined by the association equilibrium constant K A =k a / k dor dissociation equilibrium constant (also referred to as equilibrium constant) K D =k d / k a express.

[0104] Alternatively, the correlation constant K A It can be determined from equation (3), where dR / dt = 0 at equilibrium, giving:

[0105] k d R eq =k a C(R max -R eq ) (9)

[0106] where R eq is the detector response at equilibrium. Since, k a / k d =K A , substituting and rearranging in equation (9) gives:

[0107]

[0108] If the binding reaction is carried out at multiple concentrations, K A (and by extending K D ) can be obtained by nonlinear curve fitting of the data. Alternatively, for example, when the kinetic data are unreliable or the binding and dissociation are too fast to be accurately measured, R eq / C can be used to eq Plotted, this gives slope = -K A (Scatchard diagram).

[0109] Rearranging equation (10) gives:

[0110]

[0111] K A =1 / K D Substituting into equation (11) gives:

[0112]

[0113] Typically, equation (12) is modified as follows:

[0114]

[0115] Where "offset" is a compensation factor for the parallel baseline shift caused by systematic refractive index error. The offset parameter means that when the concentration C is zero, the response signal R eqDoes not need to be equal to zero. This allows the shape of the response curve to be better preserved in cases where the best fit curve does not naturally pass through zero due to baseline shift factors. If such factors are small and not a concern, the offset term can be zero.

[0116] Equations (11) and (12) can be modified by introducing a steric interference factor n, which specifies how many binding sites are blocked on average by one analyte molecule:

[0117]

[0118]

[0119] To enhance visualization of low-response data, response data can be normalized before further analysis.

[0120] Free fitting using software-assisted analysis

[0121] As described above, one conventional method for determining interaction parameters relies on recording the sensor response for various analyte concentrations and then fitting the response data to a predetermined interaction model, such as one of the models just described, without any further constraints on the parameters. In the context of the present disclosure, this approach is referred to as a "free fit" approach because the analyte saturation parameter (R) is not constrained during fitting. max or To sat ) impose no constraints. Instead, the goal is simply to find the best fit between the sensor response data and the interaction model and directly read out the output value of the analyte saturation parameter generated by this fit. Thus, from this fit, various parameters such as the saturation response R max and interaction parameter K D and K A Because the improved method of the present disclosure is built upon this existing free fitting method, the free fitting method will now be described in more detail.

[0122] Software for analyzing kinetic and affinity data is commercially available. Thus, for example, Evaluation of instrument-generated kinetic and affinity data is typically performed using the dedicated BIACORE Insight kinetics and affinity parameters by finding the parameter values ​​that give the closest fit, thereby minimizing the sum of squared residuals, as performed using kinetics software (provided by Cytiva, Uppsala, Sweden).

[0123] Thus, in one embodiment, the free fitting method involves using BIACORE Insight The software determines the affinity constant from the measured steady-state binding levels:

[0124] (i) For various analyte concentrations, the steady-state region of the curve (usually Figure 2 The reporter spots on the sensorgram in region M (when the sensorgram response is essentially parallel to the x-axis) achieve steady-state binding levels (R eq );

[0125] (ii) Manufacturing R eq a graph of its relationship to C; and

[0126] (iii) fitting the plot to an interaction model, such as a general "steady-state affinity" fitting model (e.g., Equation 13 or 14), to obtain K A / K D and R max .

[0127] Step (iii) specifically represents a "free fit" approach, where the sensor response data are fit to the interaction model (in this case Eq. 13 or 14) without any constraints on the analyte saturation parameters.

[0128] An example of this free fitting process is about Figure 3 and Figure 4 Shown. Figure 3 An example sensor response is shown, where the equilibrium response (R eq In this example, the binding behavior of the analyte is stable over a wide concentration range, and thus the data are applicable to a concentration range from approximately 10 μM to 10 mM. Figure 4 It shows that by Figure 3 The curve is obtained by fitting the data points to the interaction model (Equation 13 in the above example). The inflection point of the curve provides the K D value, which in this example is equal to approximately 0.5 mM, as indicated by the vertical dashed line. The asymptote of the fitted curve provides the R max (i.e., the maximum response value). In the example shown, R max It can be seen to be approximately 55 RU, as indicated by the horizontal dashed line.

[0129] Because in this example data are available for a wide range of analyte concentrations, Figure 4 The fit shown is reliable, and the K determined from the fit is D and R maxThe values ​​of are precise and reproducible. However, data are not always available for such a wide range of analyte concentrations. In particular, the binding behavior of many analytes becomes unstable at higher concentrations due to solubility issues. For example, some analytes will be unstable at concentrations close to K D In this case, only limited response data can be obtained at low analyte concentrations. Examples of response data for this analyte are given in Figure 5 For comparison purposes, Figure 5 Simply shows Figure 3 However, such a set of data points represents an analyte that exhibits solubility issues at higher concentrations.

[0130] Figure 6 Shows the Figure 5 The data points are fitted to the above Figure 4 The curves obtained by the same interaction model used in (i.e., the model exemplified by Equation 13) are shown. It can be seen that the K determined by fitting D and R max The value of Figure 4 The values ​​obtained in Figure 4 There are more data points available in . Now, K D It was determined to be only around 0.1 mM (see Figure 4 is 0.5 mM), and R max It can be seen that around 22.5RU (see Figure 4 In other words, in this example, without additional restrictions or constraints on the analyte saturation parameter, the response data (in this case, R max ) to the interaction model significantly underestimates K D and R max In many cases the reverse is true, and when fitting from limited data the fit overestimates K D and / or R max In either case, K D 、R max The resulting values ​​and any additional derived interaction parameters (such as K A ) are imprecise and difficult to replicate.

[0131] Therefore, it can be seen that this traditional free fitting approach is unreliable when only limited data are available, especially when the affinity of the analyte is low and data are only available below the K DThis is problematic because many useful applications and analyses (e.g., fragment-based drug design and fragment affinity analysis) involve analytes with low affinities, where high analyte concentrations are unlikely to be stable. Furthermore, even if it were theoretically possible to obtain data at sufficiently high concentrations, this could result in very expensive analyses due to the large amounts of analyte required at higher concentrations. This high cost could mean that the experiment is practically infeasible at large volumes and throughputs.

[0132] Using fixed analyte saturation parameter values

[0133] As mentioned above, one way to address this issue with free fitting methods has been suggested to be by estimating the analyte saturation response R max The method exemplified in EP2507618 relies on using a control analyte to first determine the control analyte saturation parameter R. maxC Then, the control parameter R maxC Converted to R of the analyte of interest maxA The corresponding value of R maxA can be used as a fixed value during fitting.This approach will now be briefly described.

[0134] Figure 7 A method 700 for determining a fixed value for a saturation control parameter is schematically shown in FIG. At block 702, a sensor surface is provided having a ligand immobilized thereon. This can be considered a setup or configuration step that can be performed manually or automatically. The following steps (i.e., blocks 704-716) then characterize a separate method that can be computer implemented and performed in an automated manner by a sample analysis system. At block 704, the sensor surface is contacted with a control analyte. At block 706, a sensor response from binding of the control analyte to the binding site of the ligand is recorded. At block 708, a control saturation response (R) for the interaction between the control analyte and the ligand is determined. maxC ).

[0135] At block 710, the control saturation response (R maxC ) is converted into the analyte saturation response (R maxA ). At block 712, the sensor surface is contacted with one or more samples containing different concentrations of the analyte of interest. At block 714, the sensor response from the binding of the analyte to the binding site is recorded. Finally, at block 716, the analyte saturation response (R maxA) are fitted to a predetermined interaction model (e.g., Equation 13 above) to determine the interaction parameters.

[0136] Block 702, which provides a sensor surface having a ligand immobilized thereon, involves any suitable means of immobilizing the ligand to the sensor surface in the particular biosensor being used. The immobilized ligand provides binding sites for the control analyte and the analyte to be studied. In block 704, a suitable control analyte is contacted with the sensor surface. In block 708, the control analyte should preferably be selected so that the control saturation response (R) of the interaction between the control analyte and the ligand is maxC ) can be easily determined. For example, the control analyte has a higher affinity K than the analyte(s) of interest being studied. D In one suitable example, the control analyte is provided at a concentration such that it occupies all binding sites of the ligand on the sensor surface to provide a controlled saturation response (R maxC Alternatively, the control analyte is provided in a concentration range such that it is possible to determine the control saturation response (R) from the non-steady-state interaction between the control analyte and the ligand. maxC ).

[0137] In the control saturation response (R maxC ) is converted to analyte saturation response (R maxA ) in block 710, the saturation response (R max The fact that ) represents the response when all binding sites are occupied is used for conversion. By using the same sensor surface to determine the saturation response (R) of two different analytes (e.g., analytes S and T, respectively), max ), ideally, the saturation response (R maxS ) and (R maxT ) represent the responses generated by the same number of molecules bound to the sensor surface. Therefore, the saturation response (R maxS ) and (R maxT ) gives the relative molar response contributions of the two analytes. Therefore, assuming the relative molar response contributions of the two analytes are known or can be estimated, the saturation response (R) of an analyte is max ) can be determined by the saturation response of another analyte (R max ) is calculated (estimated). This relationship is used to control the saturation response (R maxC ) is converted to analyte saturation response (R maxA ).

[0138] As mentioned above, there are many biosensors available based on many different detection technologies, so the relative molar response of two different analytes depends on the biosensor used, but the principle still applies. In one example, in block 710, the relative molar response contributions of the analyte and control analyte are approximated by the molar weight ratio between the analyte and control analyte. This approach is valid, or at least a good approximation, for any biosensor technology that directly or indirectly records the mass of molecules bound to the sensor surface. In mathematical terms, this approach is characterized by Equation 16:

[0139]

[0140] Among them, Mw C is the molar weight of the controlling analyte, and Mw A is the molar weight of the analyte of interest.

[0141] In one example, after replacing the control analyte with a sample (e.g., a buffer) that does not contain analytes capable of binding to the sensor surface, the dissociation rate of the control analyte is high enough to achieve complete dissociation within a reasonable time frame. Alternatively, a regeneration step may be required to release all binding sites between the analytes. In some examples, blocks 710 to 716 are repeated for multiple analytes to achieve complete dissociation based on the saturation response (R) calculated for each analyte. maxA ) to determine one or more interaction parameters for each of a plurality of analytes, such as Figure 7 In order to detect and / or correct for possible degradation of the sensor surface over time, steps 704-708 may be repeated after steps 710-716 have been performed for a predetermined number of analytes. Figure 7 As shown in the figure mark 720.

[0142] Figure 7 Method 700 provides an improvement over conventional free fitting methods. However, as mentioned above, the present inventors have recognized that this method is far from perfect. One disadvantage of this method is that it relies entirely on the discovery of control analytes that can serve as good substitutes for the analyte of interest. Finding good control analytes is often impossible, or at least requires a lot of experimental work. Furthermore, even when good control analytes can be found, the present inventors have recognized that the estimated interaction parameters (e.g., K) determined by this method are often not very accurate. D , K A ) is not always completely reliable. The inventors have recognized that this inaccuracy is due to Figure 7 The method relies too heavily on the fixed parameters determined (i.e., the fixed value R used in the fitting process of block 716). maxA) and caused by. R maxA Directly based on R maxC The method is therefore very dependent on maintaining the correspondence between the analyte and the control analyte, as well as determining R maxC The experiment (i.e. Figure 7 If there is imprecision in any of these aspects, i.e., the correspondence (or lack thereof) between the analyte and the control analyte and / or in determining R maxC If there is imprecision in the precision and reliability of the experiment, then this will directly feed into R maxA determination, and subsequent interaction parameters (like K D and K A ) becomes imprecise.

[0143] In view of these problems, Figure 7 Neither the free fitting method nor the more strictly controlled method can achieve satisfactorily accurate measurements in all cases. The present inventors have recognized a new method that is believed to improve the reliability of interaction parameter estimates and is independent of the control analyte. The details of this method will now be described.

[0144] New method: fitting using upper and lower limits

[0145] The present inventors have recognized that if the fitting process is constrained by an interval (i.e., upper and lower limits or upper and lower boundaries applied to the analyte saturation parameter used in the fit), then the interaction parameters can be determined more accurately. This approach effectively represents a free fitting approach (which has no constraints on the analyte saturation parameter) and Figure 7 The method (the analyte saturation parameter R maxA The inventors believe that, surprisingly, Figure 7 Having "looser" constraints on the fitting process provides a more accurate estimate of the interaction parameters than the methods of the present inventors. This is because, as the present inventors have recognized, the fitting algorithm is able to more effectively strike a balance between fitting the actual response data and taking into account known or estimated approximate values ​​of the analyte saturation parameters. The present method can be used in a more efficient way than using free fitting methods or Figure 7 No approach can balance these competing demands in an impossible way.

[0146] The present inventors have also recognized that, advantageously, the disclosed method is general and can be used both when the analyte saturation parameter is the analyte saturation response (R max ) can also be used when the analyte saturation parameter is the target occupancy at saturation (To sat) is used when the target is expressed as . This increases the versatility of the method and enables it to be applied regardless of whether the results of the experimental analysis and the associated interaction model are expressed in terms of saturation responses or target occupancy. In addition, the disclosed method does not rely on the use of control analytes (see Figure 7 ), because when control analytes are not readily available, theoretical values ​​of analyte saturation parameters can be used instead.

[0147] The disclosed novel method is schematically shown in Figure 8 middle, Figure 8 A method 800 for determining one or more interaction parameters associated with an interaction between an analyte and a ligand is described. At block 802, a sensor surface is provided with a ligand immobilized thereon. This can be considered a setup or configuration step that can be performed manually or automatically. The following steps (i.e., blocks 804-808) then represent a separate method that can be computer-implemented and can be performed in an automated manner by a sample analysis system. At block 804, the sensor surface is contacted with one or more samples containing the analyte. If only a single sample is used, it is preferred that the sample have a concentration equal to at least twice the K of the analyte. D , to enable an efficient fitting process. This is particularly true for most proteins with high affinity (low K D ) is possible because, in this case, twice the K D Still a manageable low concentration.

[0148] That is, preferably, block 804 involves contacting the sensor surface with three or more analyte samples, each sample having a different analyte concentration. This is beneficial in terms of improving reliability and because low affinity (high K) analytes may be studied more effectively. D ) interactions, where it is difficult or impossible to achieve close to (let alone twice) the K D analyte concentrations. In the example where three or more analyte samples are provided, block 804 involves contacting at least three concentrations of the analyte (provided in the three or more samples) with the sensor surface to generate concentration series data. The inventors have recognized that using 5 to 11 concentrations (provided in the 5 to 11 samples) is advantageous in balancing the need to generate useful data with practical considerations and cost.

[0149] At block 806, a sensor response indicative of binding of the analyte to the binding site of the ligand is recorded. In other words, the sensor response is recorded for each analyte concentration. Typically, for each sample that has contacted the sensor, the recorded response is the maximum response, i.e., the equilibrium (steady-state) response, which can be obtained (read out) when the sensorgram response at that concentration is substantially parallel to the x-axis. The considerations and implementation details discussed above in the general discussion of sensors and analyte-ligand interactions, as well as the discussion of the sensor response at that concentration, are discussed in detail. Figure 7 Comments related to blocks 702 through 706 of method 800 also apply to Figure 8 802 to 806 of FIG.

[0150] At block 808, the recorded sensor responses are fitted to an interaction model (e.g., Equation 13 above) to determine one or more interaction parameters associated with the interaction between the analyte and the ligand. In accordance with the presently disclosed invention, the fit is subject to predetermined upper and lower limits applied to the analyte saturation parameter associated with the interaction model. As described above, when determining an interaction parameter such as K D and K A This is believed to provide improved accuracy. The fitting process of box 808 is similar to the fitting process of box 716, and the same considerations apply, except that the fitting of box 808 is affected by, for example, R max or To sat The method may include the additional step of receiving the upper and lower limits, such as user input, for the analyte saturation parameter, and not using fixed values. It will be appreciated that the method may include the additional step of receiving the upper and lower limits, such as user input. This step may occur in Figure 8 at any point in the process.

[0151] In one implementation, the fitting process of block 808 involves applying a best fit estimate (e.g., a minimum chi-square estimate) to the recorded sensor responses within the constraints set by the interaction model and the upper and lower limits. For example, where the interaction model is represented by Equations 12 or 13 above, then the response data represents the equilibrium response at each analyte concentration, and the fitting process seeks a curve that best fits the response data. In other words, response data showing the response at equilibrium (on the y-axis) versus concentration (on the x-axis) can be plotted and fitted to Equations 12 or 13 above. In more detail, R max and K D The values ​​of can be iteratively "guessed" and each set of guessed values ​​can be fitted to the interaction model expressed in the preceding equation. max ) set the upper and lower limits, the fitting process is only allowed to consider R maxThe fit of the test values ​​within the upper and lower limits that have been set. The quality of fit (or "closeness") of each set of guessed values ​​can then be determined, for example, using a chi-square estimate as described. The best fit value can then be determined as the best estimate of the interaction parameter in question.

[0152] As mentioned above, the method of the present disclosure is not limited to R-based max In some cases, the analyte saturation response can be alternatively expressed as the target occupancy at saturation, To sat As mentioned above, To sat When R A =R max And R max with To A The target occupancy is related by the following formula:

[0153]

[0154] Similarly, the interaction model that can be fitted to the data can also be expressed in terms of target occupancy as:

[0155]

[0156] When using this interaction model, data showing the response of target occupancy (on the y-axis) to concentration (on the x-axis) can be obtained according to the method of the present disclosure based on Equation 18 and applied to To sat The upper and lower limits are used to plot and fit.

[0157] As briefly explained above, the upper and lower limits for constrained fitting can be set in a variety of ways. In one example, the system receives user input indicating the upper and lower limits. The user input can take a variety of forms. In one implementation, the interaction model used for fitting can be based on R max The biosensor system can be configured to record response values ​​in RU for different concentrations of the analyte. The user can then set the upper and lower limits as point (e.g., integer) values ​​in RU. For example, the user may know the R of the analyte of interest from past experience. max Probably around 20RU. For example, the user could set the upper limit to a point value of 25RU and the lower limit to a point value of 15RU. The same applies if the system is configured based on a target occupancy, as the user can enter the upper and lower limits based on the target occupancy, for example, an upper limit of 110% (or 1.1 in decimal) and a lower limit of 90% (or 0.9 in decimal).

[0158] Alternatively, the system can help the user set upper and lower limits by providing initial default values ​​for the analyte saturation parameters. In this case, the user input determines how far the fitting process is allowed to deviate from the default parameter values. In other words, the user specifies a value below the default parameter (i.e., the lower limit) and a value above the default parameter (i.e., the upper limit), and these values ​​then constrain the fitting in the manner described above. The upper and lower limits can be defined in terms of unit values ​​(e.g., + / - a certain number of RU units) or percentages. For example, consider a case where the system is configured to set the upper and lower limits based on R max Operation, and R max The default value is preset to 20RU. The user input can then indicate upper and lower limits relative to this default value. For example, the user can set the upper limit to +2RU and the lower limit to -2RU. In this example, the fitting process is then constrained to R max For a fit between 22 and 18 RU. Alternatively, the input can be provided in the form of percentages. For example, the user can set the upper limit to 120% and the lower limit to 80%. In this example, the fitting process is then constrained to an R between 24 and 16 RU. max These values ​​represent 120% and 80% of the default value of 20RU, respectively. A similar approach can be applied when the system operates based on target occupancy.

[0159] In implementations that use a default analyte saturation parameter as a starting point around which to set upper and lower limits, a variety of mechanisms can be utilized to determine the default value. Two primary approaches are available for determining default parameter values: theoretical estimation and experimental estimation. Either approach can be used to determine the default starting point for the analyte saturation parameter, and both approaches are suitable for different situations, as will now be explained.

[0160] A first mechanism for determining a default value for the analyte saturation parameter is to determine a theoretical estimate. This can be done based on the molecular weights of the analyte and ligand and the level of immobilization of the ligand. Specifically, in one example, R max A theoretical estimate of can be based on the following calculation:

[0161]

[0162] Among them, Mw lig is the molecular weight of the ligand, Mw ana is the molecular weight of the analyte, and R lig is the immobilization level, also known as the ligand response.

[0163] The theoretically estimated analyte saturation parameter, R max_theoThis is then used as a default value around which the upper and lower limits are set in the manner described above. Theoretically, the benefit of determining a default or initial value for the analyte saturation parameter in this way is that no control analyte needs to be used. Only the molecular weights of the ligand and analyte, and the immobilization level, need to be determined. max_theo It can also be expressed in terms of target occupancy as described above.

[0164] A second mechanism for determining a default value for an analyte saturation parameter is to experimentally determine an estimate using a control analyte. In this case, R is first determined for the control analyte. max The control analyte values ​​are then converted to corresponding values ​​for the analyte of interest. The details of how this process can be performed are given above in conjunction with Figure 7 The determined R maxA The value can also be based on the target occupancy as described above (for example, as To satA ) to represent. Once R has been determined maxA or To satA This value will then be used as the default analyte saturation parameter value, around which the upper and lower limits are set.

[0165] In various implementations, embodiments of the present invention may provide a technique that enables the use of constraints (e.g., with constrained R max ) rather than using models with free fitting and / or constant parameters (e.g., R max ) of one or more models.

[0166] However, other alternatives may provide a possible workflow where model control is of low quality or theoretical parameters (e.g. R max ), you can set wide ranges for these parameters (for example, for R max ). For samples that do not provide good results, then this will help to avoid systematically misdetermining K D , while those samples that provide good results may have results close to those of the free fit model. In addition, if the model control is of sufficiently high quality, the parameter range can be set to be narrower than the previous wide range of parameter ranges. In this way, various such embodiments provide the advantage that the user does not have to evaluate which fitting process to use for each sample series being analyzed, but can simply set the range for the entire evaluation.

[0167] Various such ranges may be predetermined, automatically generated (e.g., dynamically generated), or user defined as appropriate. Thus, depending on the sample being analyzed, automatic software switching between various models (e.g., using free-fit, constrained, or constant parameters) may be provided to provide an optimal analysis without the user having to manually determine which model is best in a particular given situation.

[0168] For example, although when the expected R is correctly determined max The constant model will generally show better performance, especially for low responses, but the question remains about the expected R max How precisely can it be determined, since the expected R max From R max Control / input or ligand levels are determined. For example, the user may be advised to use R max Control / Input. In R max In the case of normal changes in control / input (for example, when determining R max K D When the R max Equal to or better than using a constrained model with low response. However, if max If there are any outliers in the input, then there may be >> 20% correct values, but it is preferable to use a constrained model. Therefore, it can depend on the response and / or R max Values, etc. to trigger switching between the different models used.

[0169] The above methods of the present disclosure can be advantageously implemented by a sample analysis system (e.g., a biosensor). An example sample analysis system that can be used to implement the disclosed methods is produced by Cytiva T200 instrument. Figure 9 It will be understood that this is a non-limiting example of a biosensor system and that the components described below can be generalized to other forms of biosensor systems. More generally, the disclosed method can be combined and applied to a variety of other biosensor systems operating under similar principles and is not limited to implemented in the system.

[0170] The analysis system 900 includes a sample chamber 902, which is Figure 9902 and is shown in the closed position. The sample chamber 902 can be opened and (one or more) sample solutions (typically stored in one or more sample reservoirs such as vials) can be injected into the chamber 902. Once sealed, an observation window 906 allows the sample solution in the chamber 902 to be observed from the outside. During analysis, a sample delivery system (also referred to as a fluid handling system) delivers the sample solution from the sample reservoir held in the sample chamber 902 to the sensor surface. At its simplest, the sample delivery system includes a network of at least one pump and flow channels for sending the sample and running buffer to the sensor through the analysis system 900. The sensor can be inserted and removed via the sensor chip port 904 before / after analysis. A continuously running buffer can be provided from the buffer reservoir 908 via a buffer pump (not shown) housed in the buffer pump chamber 910. Waste material can be collected in the waste reservoir 912. The method disclosed herein can be implemented by a computer system (not shown) housed in or connected to the analysis system 900.

[0171] Finally go to Figure 10 , Figure 10 A schematic and simplified representation of a computer device 1001 is shown that can be used to perform the methods described herein, either alone, in combination with other computer devices, or as part of a "cloud" computing arrangement. For example, the computer device 1001 can form part of the analysis system 900 or be connected to the analysis system 900 (wirelessly or via a wired connection) and can be configured to cause the analysis system 900 to perform the various methods disclosed herein. The computer device 1001 can also perform analysis on readings or information generated by sensors of the analysis system 900.

[0172] The computer equipment 1001 in the illustrated example includes various data processing resources, such as a processor 1002 (specifically, a hardware processor) coupled to a central bus structure. Also connected to the bus structure are other data processing resources, such as a memory 1004. A display adapter 1006 connects a display device 1008 to the bus structure. One or more user input device adapters 1010 connect user input devices 1012 such as a keyboard and / or mouse to the bus structure. One or more communication adapters 1014 are also connected to the bus structure to provide connections to other computer systems 1001 and other networks. The computer equipment 1001 can be a local computer or a server. It can be an independent component, or it can be a part for the existing computer hardware used in the sample analysis laboratory.

[0173] In operation, the processor 1002 of the computer system 1001 executes a computer program comprising computer executable instructions that may be stored in the memory 1004. When executed, the computer executable instructions may cause the computer system 1001 to cause the sample analysis system 900 to perform one or more methods described herein. Figure 2 The sensor map shown) can be displayed to the user via the display adapter 1006 and the display device 1008. User input for controlling the operation of the computer system 1001 can be received from the user input device 1012 via the user input device adapter 1010.

[0174] It will be understood that Figure 10 Some features of the computer system 1001 shown may not be present in certain circumstances. For example, one or more of the plurality of computer devices 1001 may not require a display adapter 1006 or a display device 1008. This may be the case, for example, with certain computer devices 1001 that are used solely for their processing power and do not need to display information directly to a user. Similarly, user input device adapter 1010 and user input device 1012 may not be required. In its simplest form, the computer device 1001 includes a processor 1002 and a memory 1004.

[0175] As described above, the described method can be implemented using computer-executable instructions. A computer program product or computer-readable medium may include or store computer-executable instructions. A computer program product or computer-readable medium may include a hard drive, a flash memory, a read-only memory (ROM), a CD, a DVD, a cache, a random access memory (RAM) and / or any other storage medium in which information is stored for any duration (e.g., an extended period of time, permanently, temporarily, for temporary buffering and / or for caching of information). A computer program may include computer-executable instructions. A computer-readable medium may be a tangible or non-transitory computer-readable medium. The term "computer-readable" includes "machine-readable".

[0176] Although various specific combinations of components and method steps have been described, these are merely examples. Components and method steps may be combined in any suitable arrangement or combination. Components and method steps may also be omitted to leave any suitable combination of components or method steps.

[0177] The singular terms "a" and "an" should not be understood to mean "one and only one". Instead, unless otherwise specified, they should be understood to mean "at least one" or "one or more". The word "comprise" and its derivatives including "comprising" and "containing" include each stated feature but do not exclude the inclusion of one or more additional features.

[0178] The above implementations are described by way of example only, and the described implementations are to be considered in all respects as illustrative and not restrictive. It will be understood that variations may be made to the described implementations without departing from the scope of the present disclosure. It will also be understood that there are many variations not described, but such variations fall within the scope of the appended claims.

Claims

1. A method (700) for determining one or more interaction parameters associated with an interaction between an analyte and a ligand, the method comprising: contacting a sensor surface having a ligand immobilized thereon with one or more samples containing an analyte (702); recording ( 706 ) a sensor response indicative of binding of the analyte to the binding site of the ligand; as well as fitting (716) the recorded sensor responses to an interaction model to determine one or more interaction parameters associated with the interaction between the analyte and the ligand, wherein the fitting is subject to predetermined upper and lower limits applied to an analyte saturation parameter associated with the interaction model.

2. The method (700) of claim 1, wherein: The sensor surface is contacted with three or more samples containing different concentrations of the analyte.

3. The method (700) according to claim 1 or 2, wherein: The one or more interaction parameters include one or more of the following: The association equilibrium constant K of the interaction A ;and The dissociation equilibrium constant K of the interaction D .

4. The method (700) according to any preceding claim, wherein: The analyte saturation parameter is the analyte saturation response value R max .

5. The method (700) according to any one of claims 1 to 3, wherein: The analyte saturation parameter is the target occupancy value To at saturation. sat .

6. The method (700) according to any preceding claim, further comprising: determining (708) an estimate of the analyte saturation parameter; as well as At least one of the upper and lower limits is set based on the estimated value of the analyte saturation parameter.

7. The method (700) of claim 6, wherein: At least one of the upper and lower limits is set as a percentage of the estimated value of the analyte saturation parameter.

8. The method (700) of claim 7, wherein: The lower limit is set between 75% and 99% of the estimated analyte saturation parameter.

9. The method (700) of claim 7, wherein: The lower limit is set between 25% and 99% of the estimated analyte saturation parameter.

10. The method (700) according to claim 7, 8 or 9, wherein: The upper limit is set between 101% and 150% of the estimated analyte saturation parameter.

11. The method (700) according to claim 7, 8 or 9, wherein: The upper limit is set between 101% and 250% of the estimated analyte saturation parameter.

12. The method (700) according to any one of claims 6 to 11, wherein: Determining an estimate of the analyte saturation parameter includes calculating a theoretical estimate of the value based on: The molecular weight Mw of the ligand lig ; The molecular weight Mw of the analyte ana ;as well as Ligand response R lig .

13. The method (700) according to any one of claims 6 to 11, wherein: Determining an estimate of the analyte saturation parameter includes: contacting the sensor surface with a control analyte (704); recording (706) a sensor response indicative of binding of the control analyte to the binding site of the ligand; determining (708) a control analyte saturation parameter for the control analyte; and The estimate of the analyte saturation parameter is calculated based on the determined control analyte saturation parameter and a molar weight ratio between the analyte and the control analyte.

14. The method (700) of any preceding claim, further comprising receiving user input setting the upper and lower limits.

15. The method (700) of any preceding claim, wherein: The fitting includes applying a minimum chi-square estimate to determine the closeness of the fit to the recorded sensor responses.

16. The method (700) of any preceding claim, wherein: The predetermined interaction model includes the following expression: in R eq is the sensor response recorded at equilibrium, C is the analyte concentration, R max is the analyte saturation parameter, and K D is the dissociation constant of the interaction.

17. The method (700) of claim 16, wherein: The predetermined interaction model also includes an offset term to compensate for the parallel baseline shift due to systematic refractive index errors.

18. The method (700) of any preceding claim, wherein: Recording the sensor response indicative of binding of the analyte to the binding site of the ligand is based on surface plasmon resonance (SPR).

19. The method (700) according to any one of claims 1 to 17, wherein: Recording a sensor response indicative of binding of the analyte to the binding site of the ligand is based on evanescent wave sensing.

20. A sample analysis system (900) for determining an interaction parameter, comprising: a sensing surface configured to detect a binding interaction between an analyte and a ligand at the sensing surface; as well as A computing device configured to cause the sample analysis system to perform the method of any one of claims 1-19.

21. A computer configured to cause a sample analysis system to execute the method according to any one of claims 1-19.

22. A computer-readable storage medium comprising instructions which, when executed by a computer, cause the computer to cause a sample analysis system to perform the method (700) of any one of claims 1-19.

23. A computer program comprising instructions which, when executed by a computer, cause the computer to cause a sample analysis system to perform the method (700) of any one of claims 1-19.

Images