Method, system and device for predicting PK value of antibody

Abstract

Methods, systems and devices are described for generating a pharmacokinetic (PK) model for predicting a PK value of an antibody of interest. An input data set is received for a plurality of antibodies. The input data set includes data representing an amino acid sequence of each of the antibodies and an experimentally determined PK value of each of the antibodies. One or more surface properties for each of the antibodies are calculated based on the corresponding amino acid sequence. One or more area surface properties for one or more regions of interest are calculated for each of the antibodies based on the one or more surface properties calculated for each of the antibodies. A grouping is determined from one or more of the regions of interest that results in a maximum correlation between the corresponding calculated area surface property of the grouping and the corresponding experimentally determined PK value. This is used to establish a PK-area surface property relationship for the plurality of antibodies. A PK model for predicting a PK value of the antibody of interest is generated based on the PK-area surface property relationship. The PK model is configured to receive one or more input area surface properties of the determined grouping for the antibody of interest and output a predicted PK value for the antibody of interest by applying the input area surface properties to the relationship.

Description

[Technical field]

[0001] The present specification relates to methods, systems and devices for predicting pharmacokinetic (PK) values ​​of antibodies prior to in vitro or in vivo analysis. [Background technology]

[0002] The introduction of therapeutic monoclonal antibodies (mAbs) into clinical practice has revolutionized healthcare to the point of becoming the top selling drug format over the past few years. The growing popularity of antibody-based therapeutics has led to the development of next-generation multispecific antibody therapeutics that combine two or more variable regions into a single molecule. Due to their structure, multispecific antibodies are able to engage two or more antigens at once, thereby offering new opportunities to tackle complex diseases compared to monospecific antibody combinations. To date, most of the available information is derived from monospecifics.

[0003] In recent years, the pharmaceutical industry has shown great interest in elucidating the pharmacokinetic (PK) principles of antibodies, since they provide important information regarding drug efficacy and dosing strategies. During antibody discovery / optimization campaigns, candidates with unacceptable PK characteristics are abandoned (or redesigned). Due to budgetary limitations and ethical concerns, in vivo testing cannot be performed on a large panel of candidates. Therefore, de-risking before performing in vivo PK evaluation is highly desirable.

[0004] As a result, many efforts have aimed to correlate in vivo PK data with in vitro properties and in silico descriptors. The literature describes a series of experimental assays and estimated biophysical properties of mAbs that can be used to predict the PK of monospecific antibodies. These include, for example, polyspecific reagent binding assays, affinity capture self-interaction nanoparticle spectroscopy, binding to neonatal Fc receptors (FcRn), and computational estimation of antibody surface properties. It has been reported that mAbs with excessively positive surfaces are more frequently internalized into cells by nonspecific pinocytosis and exhibit increased binding affinity for FcRn, thus preventing the dissociation of the complex. Thus, the positive charge may interfere with the IgG recycling pathway and negatively affect PK. However, since this hypothesis is derived from conventional mAbs, its significance for multispecific formats remains unclear. Summary of the Invention [Problem to be solved by the invention]

[0005] Therefore, there is a need for an efficient, accurate and rapid mechanism or device for predicting the PK value of any antibody of interest, which allows for the short-listing of multiple antibodies that have not been tested or have unknown PK values. In particular, there is a need for an efficient, accurate and rapid mechanism or device for predicting the PK value of a multispecific antibody of interest, which allows for the short-listing of multiple multispecific antibodies that have not been tested or have unknown PK values. [Means for solving the problem]

[0006] According to a first aspect of the present specification, there is provided a computer-implemented method for generating a pharmacokinetic (PK) model for predicting a PK value for an antibody of interest, the method comprising: receiving an input dataset for a plurality of antibodies, the input dataset comprising data representing an amino acid sequence of each of the antibodies and an experimentally determined PK value for each of the antibodies; calculating one or more surface properties for each of the antibodies based on the corresponding amino acid sequences; calculating one or more region surface properties for one or more regions of interest for each of the antibodies based on the one or more surface properties calculated for each of the antibodies; determining from the one or more regions of interest a grouping that results in a maximum correlation between the corresponding calculated region surface property of the grouping and the corresponding experimentally determined PK value to establish a PK region surface property relationship for the plurality of antibodies; and generating a PK model for predicting a PK value for the antibody of interest, the PK model being configured to receive one or more input region surface properties of the determined grouping for the antibody of interest and output a predicted PK value for the antibody of interest by applying the input region surface property to the PK region surface property relationship.

[0007] The computer-implemented method may further include a PK model further configured to calculate one or more surface properties for the antibody of interest based on its amino acid sequence; and calculating one or more region surface properties for the grouping of regions of interest determined for the antibody of interest based on the one or more surface properties calculated for the antibody of interest, thereby providing said input region surface properties.

[0008] The computer-implemented method, wherein the generated PK model is further configured to predict a shortlist of candidate antibodies having desired PK properties for use in in vitro wet lab analysis or in in vivo testing.

[0009] The computer-implemented method further comprises the steps of: for each of the candidate antibodies, calculating one or more surface properties for each of the candidate antibodies based on the corresponding amino acid sequence; calculating one or more area surface properties for the groupings determined from the areas of interest for each of the candidate antibodies; inputting data representing the calculated area surface properties for each of the candidate antibodies into a generated PK model to predict a PK value for each of the candidate antibodies; receiving the predicted PK values ​​for each of the candidate antibodies as output from the PK model; adding the candidate antibody to a shortlist of candidate antibodies if the received predicted PK value of the candidate antibody exhibits one or more of the desired PK properties defined for the shortlist; and outputting data representing the shortlist of candidate antibodies for use in at least in in vivo testing or in vitro wet lab analysis.

[0010] A computer-implemented method, wherein calculating one or more surface properties for each of a plurality of antibodies comprises modeling a three-dimensional molecular structure of each of the antibodies, and calculating a distribution for the one or more surface properties across a surface of the modeled molecular structure of each of the antibodies.

[0011] A computer-implemented method, wherein the one or more calculated surface properties include one or more of a positively charged surface area, a negatively charged surface area, and a hydrophobic surface area.

[0012] A computer-implemented method, wherein the surface of the modeled molecular structure of each of the antibodies comprises a plurality of patches of one or more surface properties, each patch having a patch area based on the distribution of the surface properties in the modeled molecular structure, and the number of patches for each antibody is the same or different for each other antibody of the plurality of antibodies.

[0013] A computer-implemented method, wherein calculating one or more area surface properties for one or more regions of interest for each of said antibodies is based on an area of ​​said patch of one or more surface properties associated with each region of interest.

[0014] A computer-implemented method, wherein the plurality of antibodies and the antibody of interest is a crossover dual variable (CODV) antibody.

[0015] A computer-implemented method, in which at least one region of interest is selected from a variable heavy chain, a VH domain or a variable light chain, a complementary domain region of a VL domain, a CDR, a framework region or a linker, including CDR1, CDR2, CDR3, FW1, FW2, FW3, FW4 of either the VH domain or the VL domain of the two variable V domains of a CODV antibody.

[0016] A computer-implemented method, wherein the grouping from the region of interest is CDR1 of VL1, CDR3 of VL1 and FW1-4 of VL2 and VH2 of the CODV antibody.

[0017] The computer-implemented method, wherein the PK surface property relationship is a linear PK surface property relationship.

[0018] According to a second aspect of the present specification, there is provided an apparatus comprising a processor, a memory unit and a communication interface, the processor being connected to the memory unit and the communication interface, the processor and the memory being configured to perform a computer-implemented method according to any of the features or steps of the first aspect.

[0019] According to a third aspect of the present specification, there is provided a computer readable medium comprising data or instruction code which, when executed on a processor, causes the processor to perform a computer implemented method of any of the features or steps of the first aspect.

[0020] According to a fourth aspect, there is provided a system comprising: a three-dimensional surface property module configured to receive data representing a plurality of candidate antibodies and calculate area surface properties of a region of interest for each of the candidate antibodies; a pharmacokinetic (PK) model module configured to receive the calculated area surface properties corresponding to each candidate antibody for predicting a PK value for each of said candidate antibodies; a PK comparison module configured to compare the predicted PK values ​​of the candidate antibodies with desired PK properties and select a candidate antibody from the plurality of candidate antibodies having a PK value that satisfies the one or more desired PK properties; and an output module configured to output a shortlist of candidate antibodies from the selected candidate antibodies based on the comparison for use in in vitro wet lab analysis and / or in vivo testing.

[0021] The system may be further configured to perform one or more of the method steps or features according to the first aspect.

[0022] According to a fifth aspect of the present specification, there is provided a non-transitory tangible computer readable medium comprising data or instruction codes for generating a PK model for predicting a PK value for an antibody of interest, the PK model, when executed on one or more processors, comprising the steps of: receiving an input dataset for a plurality of antibodies, the input dataset comprising data representing an amino acid sequence of each of the antibodies and an experimentally determined PK value for each of the antibodies; calculating one or more surface properties for each of the antibodies based on the corresponding amino acid sequence; and calculating one or more surface properties for one or more regions of interest for each of the antibodies based on the one or more surface properties calculated for each of the antibodies. a step of calculating region surface properties above; a step of determining, from one or more regions of interest, a grouping that results in a maximum correlation between the corresponding calculated region surface properties of the grouping and the corresponding experimentally determined PK value to establish a PK region surface property relationship for a plurality of antibodies; and a step of generating a PK model for predicting a PK value for an antibody of interest, wherein the PK model is configured to receive one or more input region surface properties of the determined grouping for the antibody of interest and output a predicted PK value for the antibody of interest by applying the input region surface properties to the PK region surface property relationship.

[0023] The non-transitory tangible computer readable medium may be further configured to perform one or more of the method steps or features according to the first aspect.

[0024] In order that the invention may be more easily understood, embodiments thereof will now be described, by way of example only, with reference to the accompanying drawings in which: [Brief description of the drawings]

[0025] [Figure 1a] 1 illustrates an exemplary PK model pipeline process according to some embodiments of the present invention. [Figure 1b]1 illustrates an exemplary PK model process according to some embodiments of the present invention. [Figure 1c] 1 illustrates an exemplary antibody shortlisting process according to some embodiments of the present invention. [Figure 2a] 1 shows an exemplary CODV antibody structure and an exemplary three-dimensional (3D) model of a CODV antibody according to some embodiments of the present invention. [Figure 2b] 1 shows further examples of CODV antibody structures according to some embodiments of the present invention. [Figure 2c] 1 shows an exemplary multispecific CODV Ig-like dataset according to some embodiments of the present invention. [Figure 3a] 1 illustrates a system 300 for generating and using a PK model according to some embodiments of the present invention. [Figure 3b] 1 illustrates an in silico computational pipeline according to some embodiments of the present invention. [Figure 3c] 1 shows regions of interest of CODV antibody structures according to some embodiments of the present invention. [Figure 4a] 1 shows a series of plots of correlations between experimental PK clearance values ​​and ionic properties in two regions of interest, according to some embodiments of the present invention. [Figure 4b] 1 shows a series of plots of the correlation between experimental PK clearance values ​​and hydrophobicity in regions of interest, according to some embodiments of the present invention. [Figure 4c] 4b shows a cluster plot of the relationship between the plots of FIG. 4a and FIG. 4b according to some embodiments of the present invention. [Figure 4d] FIG. 4B shows a correlation plot of the combined scores (shown as in silico clearance likelihood) of the plots of FIG. 4a and FIG. 4b, according to some embodiments of the present invention. [Diagram 5] 1 shows a plot relating to two test CODV antibodies where both test antibodies contain the same V domain but their orientations are swapped, the plot is the ratio of positive ions in CDR1 and CDR3 of VL1 relative to the hydrophobic region in the FW of VR2. [Figure 6] Shown are plots relating to two test CODV antibodies where both test antibodies contain the same V domains but their orientations are swapped, the plot being the positive ion ratio in CDR1 and CDR3 of VL1 relative to the hydrophobic region in the FW of VR2 (left) and the combined score (referred to as in silico clearance likelihood) plotted against the experimentally determined in vivo clearance (right). [Figure 7] FIG. 1 shows a correlation plot of the combined scores (referred to as in silico clearance likelihood) of a validation set of seven new bispecific CODV molecules that were first predicted for their PK properties using the PK model pipeline of the present invention and then evaluated for in vivo PK (clearance). [Figure 8] FIG. 1 is a schematic diagram of a system / apparatus for carrying out the methods described herein. DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS

[0026] Common reference numbers are used throughout the drawings to denote similar features.

[0027] Various exemplary embodiments described herein relate to method(s), apparatus, and system(s) for automated, efficient, and reliable in silico prediction or estimation of pharmacokinetic (PK) values ​​of antibodies using PK models generated from experimental PK data of a set of antibodies. PK values ​​of interest include, but are not limited to, for example, PK clearance or clearance (in mL / h / kg), PK half-life (T1 / 2), and PK volume of distribution (Vss) and / or any other suitable PK value of interest. The PK models are generated from a computational pipeline that estimates the surface patch landscape of an antibody (e.g., a crossover dual variable (CODV) antibody and / or any other type of antibody), analyzes the landscape with experimental PK data available for a set of known antibodies, and identifies groupings of regions of interest on the landscape that are maximally correlated with the experimental PK data to form PK value relationships.

[0028] A PK model is formed from the identified grouping of PK value relationships and regions of interest. Once formed, the PK model can be used to identify candidate antibodies of interest and generate a shortlist of antibodies that meet specific PK value requirements (e.g., fast or slow in vivo clearance) prior to in vitro wet lab analysis and / or in vivo testing of the shortlisted antibodies. The surface patch landscape includes a plurality of surface patches (e.g., regions on the surface of the antibody that have specific surface properties), each having a surface patch property including, but not limited to, at least one of ionic properties and hydrophobicity.

[0029] The region of interest of an antibody may include, but is not limited to, for example, one or more portions of an antigen-binding fragment (Fab) region, a variable binding region of a Fab; portions of a variable binding region associated with variable domain 1 (VR1) or 2 (VR2); complementarity determining regions (CDRs) corresponding to each of the VR1 and VR2 domains of an antibody; framework (FW) regions corresponding to each of the VR1 and VR2 domains of an antibody; one or more portions of a variable heavy chain (VH) region and / or a variable light chain (VL) binding region of a VR1 and / or VR2 binding region of an antibody; one or more portions of a VH1, VH2, VL1 and VL2 binding region of a VR1 and / or VR2 binding region of an antibody; CDR regions and FW regions associated with each of the VH1, VH2, VL1 and VL2 binding regions of an antibody; one or more linkers (L1-L4) of an antibody molecule; and / or any other combination of regions of interest of an antibody as required by the application.

[0030] The embodiments of the PK models and / or processes described herein provide the advantage of efficient and accurate prediction and estimation of PK values ​​(e.g., PK clearance) in silico, rather than performing expensive and laborious in vitro and / or in vivo testing. An additional advantage of the PK models described herein is the reduction in the number of candidate antibodies and / or CODV antibodies that may have unknown PK, and the ability to efficiently shortlist candidate antibodies or CODV antibodies such that only the most promising candidate antibodies or CODV antibodies are selected for in vitro wet lab analysis or screening and / or in vivo testing. Thus, only antibodies with estimated / predicted PK values / characteristics (e.g., PK clearance) that meet the desired or required PK thresholds or characteristics can be shortlisted prior to expensive and / or laborious in vitro wet lab analysis and / or subsequent in vivo testing.

[0031] The method of the present application allows for the first time robust PK prediction of multispecific, e.g. bispecific, antibodies. Furthermore, we successfully correlate the size of the patch of surface properties with the PK value. For CODV antibodies, it has been established that the orientation of V1 and V2 affects the PK value.

[0032] 1a shows an exemplary PK model pipeline process 100 for generating a PK model for predicting the PK value of an antibody of interest. The PK model pipeline process 100 includes at least the following steps:

[0033] In step 101, an input dataset for a plurality of antibodies is received, the input dataset including data representing the amino acid sequence of each of the plurality of antibodies and an experimentally determined PK value (eg, PK clearance) for each said antibody.

[0034] In some embodiments of the invention, the plurality of antibodies and the antibody of interest is a multispecific antibody.

[0035] In some embodiments of the invention, the plurality of antibodies and the antibody of interest is a CODV antibody. In some embodiments of the invention, the plurality of antibodies and the antibody of interest is a tetravalent, bispecific CODV antibody. In some embodiments of the invention, the plurality of antibodies and the antibody of interest is a trivalent, trispecific CODV antibody. In some embodiments of the invention, the plurality of antibodies and the antibody of interest is a bivalent, bispecific CODV antibody.

[0036] In step 102, one or more surface properties are calculated for each of the plurality of antibodies based on their corresponding amino acid sequences.

[0037] For example, calculating one or more surface properties for each of a plurality of antibodies may include, but is not limited to, modeling a three-dimensional (3D) molecular structure of each of the antibodies and calculating a distribution of the one or more surface properties on the surface of the 3D modeled molecular structure of each of the antibodies.

[0038] Essentially, several types of surface properties can be calculated for each of the antibodies, including, but not limited to, ionic surface properties or hydrophobic surface properties. Ionic surface properties include negatively charged surface areas and positively charged surface areas. Hydrophobic surface properties include hydrophobic surface areas. Thus, the calculated one or more surface properties of each antibody include a surface property area based on one or more of the positively charged surface area, the negatively charged surface area, and the hydrophobic surface area on the surface of the 3D molecular structure of each antibody. The modeling of the 3D structure and the distribution of the one or more surface properties on the surface of the 3D structure can be repeated several times for each of the antibodies. The results of these repetitions can then be averaged. In one embodiment, the modeling of the 3D structure and the distribution of the one or more surface properties on the surface of the 3D structure are repeated N times (e.g., N=50, N=100, or any other suitable sample size) and then averaged.

[0039] The surface of the modeled molecular structure of each of the antibodies comprises one or more surface property regions of one or more surface properties, each surface property region comprising a plurality of patches of one or more surface properties, each patch having a patch area based on the distribution of the surface properties in the modeled molecular structure, the number of patches for each antibody being the same or different for each antibody of the plurality of antibodies, each patch being a negatively charged patch; a positively charged patch; or a hydrophobic patch.

[0040] In another example, the surface feature of at least one of the multiple patches on the surface of each antibody comprises an ionic surface feature, and the ionic surface feature of each patch comprises a positively charged ionic region or a negatively charged ionic region.

[0041] In step 103, one or more regional surface properties are calculated for one or more regions of interest for each of said antibodies based on the one or more surface properties calculated in step 102 for each of said antibodies.

[0042] For example, each region of interest on the surface of the modeled molecular structure of each of the antibodies comprises one or more surface areas having one or more surface area properties, and each surface property region comprises one or more patches or patches. Calculating one or more region surface properties for one or more regions of interest for each of the antibodies is based on the area of ​​the patches of one or more surface properties. Each region of interest may include at least three types of surface properties, including, for example, but not limited to, negatively charged regions, positively charged regions, and hydrophobic regions.

[0043] In another example, calculating at least one regional surface property corresponding to one or more regions of interest for each of the antibodies may further include calculating, for each antibody, at least one regional surface property for each region of interest based on identifying regions of each antibody associated with each region of interest and combining corresponding surface properties of the identified regions of each antibody.

[0044] In another example, the at least one region surface property corresponding to one or more regions of interest includes an ionic surface property corresponding to one or more regions of interest. The ionic surface property of each region of interest may include a positive ionic region surface property and a negative ionic region surface property, where calculating the region surface property of each region of interest of each of the antibodies further includes calculating the region surface property of each region of interest of each of the antibodies based on: calculating the positive ionic region surface property of each region of interest of each of the antibodies by aggregating or averaging patches identified as associated with each region of interest having positively charged ionic regions; and calculating the negative ionic region surface property of each region of interest of each of the antibodies by aggregating or averaging patches identified as associated with each region of interest having negatively charged ionic regions.

[0045] In step 104, a grouping is determined from one or more regions of interest that results in a maximum correlation between the grouping's corresponding calculated region surface properties and the corresponding experimentally determined PK values. The correlations can be analyzed to establish PK region surface property relationships for multiple antibodies.

[0046] For example, a linear regression may be performed for each grouping from one or more regions of interest for all of the multiple antibodies based on plotting data points corresponding to the calculated region surface properties of the grouping for each antibody with the corresponding experimentally determined PK value for each antibody, and performing a linear regression analysis on the resulting data points.The grouping with the highest correlation based on the linear regression (e.g., the highest positive or negative correlation) is then selected as the determined grouping.The linear regression output of the selected grouping is used to establish the PK region surface property relationship of the multiple antibodies.

[0047] Determining a grouping from one or more regions of interest may further include determining a grouping based on a combination of one or more regions of interest of the antibody that results in a maximum correlation (e.g., a positive or negative correlation) between the area surface characteristics of the combination of regions of interest of the antibody and the corresponding experimentally determined PK value of the antibody. Establishing a PK surface characteristic relationship may be based on estimating a correlation between the determined grouping and the corresponding experimentally determined PK value of the antibody. For example, establishing a PK surface characteristic relationship may include calculating a linear PK surface area characteristic relationship from correlating the combined area surface characteristics of the groupings determined for each of the multiple antibodies with the corresponding experimentally determined PK value of each of the multiple antibodies.

[0048] Alternatively or additionally, a non-linear regression may be performed for each grouping from one or more regions of interest for all of the multiple antibodies based on plotting data points corresponding to the calculated region surface properties of the grouping for each antibody with the corresponding experimentally determined PK value for each antibody, and performing a non-linear regression analysis on the resulting data points.The grouping with the highest correlation based on the linear regression is then selected as the determined grouping.The linear regression output of the selected grouping is used to establish the PK region surface property relationship of the multiple antibodies.In either case, a PK region surface property relationship is established.

[0049] In step 105, generating a PK model for predicting a PK value of the antibody of interest is based on the PK region surface feature relationship of the determined grouping associated with the maximum correlation. The PK model is configured to predict a PK value of the antibody of interest by receiving one or more input region surface features of the determined grouping for the antibody of interest, processing the input region surface features of the determined grouping for the antibody of interest according to the PK region surface feature relationship, and outputting a predicted PK value of the antibody of interest. Processing may include applying said input region surface features to said PK region surface feature relationship.

[0050] The PK model may be further configured to receive data representing an antibody of interest, such as its amino acid sequence, and calculate one or more surface properties of the antibody of interest based on the amino acid sequence. The PK model may then calculate one or more region surface properties for the groupings determined from the regions of interest for the antibody of interest based on the one or more surface properties calculated for said antibody of interest, thereby providing said input region surface properties.

[0051] The PK model may also be configured to assign different PK indicators to two or more non-overlapping PK value ranges. The PK model may be further configured to determine which of two or more PK ranges the predicted PK value falls into and output a corresponding PK indicator and / or predicted PK value. For example, if the PK value is a PK clearance, the indicator may correspond to, for example, but not limited to, "slow," "intermediate," or "fast" PK clearance, or any other PK clearance rate, etc.

[0052] As an example, the generated PK model is further configured and used to predict a shortlist of candidate antibodies with desired PK properties for use in in vitro wet lab analysis and / or in vivo testing. Each candidate antibody is input into the PK model as an antibody of interest, e.g., the amino acid sequence of the candidate antibody is input, and group region surface properties are calculated based on the amino acid sequence of the candidate antibody and applied to the PK region surface property relationships. A candidate antibody exhibits desired PK properties if the estimated PK value is above or below a particular PK threshold (depending on what is desired) or if the PK value associated with the group region surface property cluster is in the PK value region associated with the shortlisted candidate antibody.

[0053] Figure 1b shows an exemplary PK model process 110 for predicting the PK value of an antibody of interest. Assume that a PK model has been generated according to the PK model pipeline process 100 of Figure 1a. The PK model process 110 includes the following steps:

[0054] In step 111, data describing the antibody of interest, such as its amino acid sequence, is received.

[0055] In step 112, one or more surface properties are calculated for the antibody of interest based on its amino acid sequence. Calculating one or more surface properties of the antibody of interest based on its amino acid sequence may further include modeling a 3D molecular structure of said antibody of interest based on its amino acid sequence, and calculating a distribution of said one or more surface properties on the surface of the modeled molecular structure of said antibody of interest.

[0056] In step 113, the determined grouping of regions of interest from step 104 is used to calculate one or more region surface properties for the antibody of interest based on the calculated one or more surface properties for the antibody of interest, thereby providing input of region surface properties for application to the PK region surface property relationship to estimate the PK value.

[0057] In step 114, the PK model generated in step 105 is used to estimate the PK value of the antibody of interest based on the calculated area surface properties by inputting data representing the calculated area surface properties into the generated PK model for predicting the PK value of each of the antibodies of interest.

[0058] In step 115, an indication of the PK value of the antibody of interest based on the estimated PK value is output.

[0059] Figure 1c shows an exemplary antibody shortlisting process 120 that uses the PK model process 110 to predict a candidate shortlist of antibodies having a desired range of PK values ​​or according to desired PK features / characteristics. It is assumed that the PK models have been generated according to the PK model pipeline process 100 of Figure 1a. The antibody shortlisting process 120 includes the following steps:

[0060] In step 121, data representing a plurality of antibodies having unknown PK values ​​is received. The data representing each antibody may include, for example, but is not limited to, the amino acid sequence of the antibody or other standard molecular structure representing the antibody.

[0061] In step 122, surface properties of each antibody are calculated based on its amino acid sequence or other standard molecular structure representing the antibody. For example, surface properties of each antibody can be calculated based on performing N 3D simulations for each antibody to calculate surface properties of each antibody based on its amino acid sequence or other standard molecular structure representing the antibody. Each 3D simulation for the antibody generates a set of surface properties for the antibody, and the N sets of surface properties generated from the 3D simulations for each antibody are aggregated and averaged.

[0062] In step 123, each antibody of the plurality of antibodies is processed to determine whether the antibody meets the PK criteria / characteristics for inclusion in the shortlist. Each candidate antibody from the plurality of antibodies is selected for testing / evaluation based on the following steps for each antibody:

[0063] In step 124, the calculated area surface properties associated with the candidate antibody selected as the antibody of interest are applied to a PK model (e.g., a PK model output by process 100 or PK model 110 of FIG. 1a or FIG. 1b) configured to predict or estimate the PK value of the antibody.

[0064] In step 125, data representing predicted or estimated PK values ​​for the antibody of interest output from the PK model is received.

[0065] At step 126, it is determined whether the outputted received PK value of the antibody of interest is within the desired range of PK values ​​required to shortlist the respective antibody. If the received PK value is within the desired range of PK values ​​or is above the desired minimum PK threshold or below the desired PK maximum threshold (e.g., Y), the antibody of interest meets the PK requirements and process 120 proceeds to step 128. If the received PK value is outside the desired range of PK values ​​or is below the desired minimum PK threshold or above the desired PK maximum threshold (e.g., Y), the antibody of interest does not meet the PK requirements and process 120 proceeds to step 127.

[0066] In step 127, the next candidate antibody in the plurality of candidate antibodies is selected as the antibody of interest, and process 120 proceeds to step 124 to test the selected antibody of interest.

[0067] In step 128, if the received predicted PK value of the candidate antibody exhibits one or more of the desired PK characteristics / ranges / thresholds defined for the shortlist, the antibody of interest is added as a candidate antibody to the shortlist of candidate antibodies.

[0068] Once all candidate antibodies in the plurality of antibodies have been processed, data is output representing a shortlist of candidate antibodies for use in at least in vitro wet lab analysis and / or in vivo testing for efficacy and / or PK analysis of said type of antibodies.

[0069] Each of the processes 100, 110, 120 of Figures 1a-1c may be implemented on an apparatus including a processor, a memory unit, and a communication interface. The processor is connected to the memory unit and the communication interface. The processor and memory may be configured to implement each of the processes 100, 110, 120 and / or other processes described herein as computer-implemented methods. The processes 100, 110, 120 and / or combinations thereof, variations thereof based on one or more other processes described herein may be stored on a computer-readable medium. The computer-readable medium may include data or instruction code that, when executed on a processor, causes the processor to implement one or more of the processes 100, 110, 120 and / or combinations thereof, variations thereof as one or more computer-implemented methods as described herein.

[0070] In some embodiments, the multiple antibodies and the antibody of interest are crossover dual variable (CODV) antibodies. However, any type of antibody may be used. It may be preferable that the multiple antibodies used to generate the PK model are of the same antibody family or type. It may be preferable that the antibody format of the multiple antibodies used to generate the PK model is the same antibody format. For example, only CODV antibodies are used to generate the CODV PK model, and only monoclonal antibodies (mAbs) are used to generate the mAb PK model. For example, only multispecific antibodies are used to generate the multispecific PK model, and only monospecific antibodies are used to generate the monospecific PK model.

[0071] "CODV antibodies" as used herein include any antibody or antibody fragment that contains two V domains with crossed orientation. CODV antibodies have been previously described in International Patent Applications WO 2012 / 135345 and WO 20161 / 16626 and the publication Steinmetz et al., MAbs. 2016 Jul;8(5):867-78, which are incorporated herein by reference. In some embodiments, the CODV antibodies are in CODV-Ig format, i.e., contain an Fc domain. In some embodiments, the CODV antibodies are in CODV-Fc-OL format, i.e., contain an Fc domain, but only one CODV-Fab. In some embodiments, the CODV antibodies are in CODV format, i.e., contain one CODV-Fab and one conventional Fab, resulting in a trispecific construct.

[0072] The CODV antibody is multispecific. As used herein, "multispecific" refers to an antibody that specifically binds to two or more targets. In some embodiments, the CODV antibody is bivalent and bispecific. In some embodiments, the CODV antibody is trivalent and bispecific. In some embodiments, the CODV antibody is trivalent and trispecific. In some embodiments, the CODV antibody is tetravalent and bispecific. In some embodiments, the CODV antibody is tetravalent and trispecific. In some embodiments, the CODV antibody is tetravalent and tetraspecific.

[0073] In some embodiments, in the case of a CODV antibody, the region of interest may be selected from CDR1, CDR2, CDR3, FW1, FW2, FW3, FW4 of either the VH domain or the VL domain of the two V domains of the CODV antibody. The grouping from the regions of interest that results in the greatest correlation may be found to be CDR1 of VL1, CDR3 of VL1 and FW1-4 of VL2 and VH2 of the CODV antibody.

[0074] Although the following description describes a PK model generation pipeline and application in the context of the clearance (or PK clearance) of a crossover dual variable (CODV) Ig-like format antibody, it should be understood by one of skill in the art that this is merely exemplary and the invention is not so limited, and the principles described herein in relation to the generation of a PK model can be applied to any type of antibody as the application requires.

[0075] 2a and 2b are exemplary CODV antibody architectures / structures 200, 210, 220, 230 and 240 according to some embodiments. The CODV antibody architecture 200 in FIG. 2a is a schematic diagram of a CODV antibody showing two VR binding domains (VR1, VR2) linked via a linker (L). VR1 and VR2 are paired together in a crossover manner, with VH1 of VR1 binding to VH2 of VR2 via L3 and VL1 of VR1 binding to VL2 of VR2 via L1. VL1 of VR1 binding to the constant domain of the light chain (CL) via L2 and VH2 of VR2 binding to the first constant domain of the heavy chain (CH1) via L4. VH1 pairs with VL1 (VR1) and VH2 pairs with VL2 (VR2). The CODV antibody architecture 210 is a 3D molecular model of a CODV molecule having VR1 and VR2 regions generated using, for example, Molecular Operating Environment (MOE), (Molecular Operating Environment (MOE), 2020.09 Chemical Computing Group ULC, 1010 Sherbooke St. West, Suite #910, Montreal, QC, Canada, H3A 2R7, 2022) or other 3D modeling software.

[0076] CODV antibody architecture 220 in FIG. 2b is another exemplary CODV architecture that includes two VR-binding domains paired together in a cross-over fashion as fragment antigen-binding (Fab) arms of immunoglobulin G (IgG). This allows the design of bispecific (bi-Ab) and trispecific (tri-Ab) modalities, shown as CODV antibody architectures 220, 230, and 240. The three modalities 220, 230, and 240 include at least one CODV-Fab arm, with the identity of the second arm differing from each other. With respect to the bispecific modalities, CODV-IgG has two identical CODV-Fab arms, while CODV-FcOL lacks a second arm and is asymmetric. The second arm of the shown CODV trispecific scaffold is a standard Fab.

[0077] In the basic structure of a CODV-Fab arm, the linear sequences of the heavy (HC) and light (LC) chains code for their respective contributions in two VR domains of the molecule, called VR1 and VR2. Each VR targets an epitope of interest, allowing for multispecificity. In terms of its structure, the CODV-Fab arm is formed by an inner domain and an outer domain connected by two linkers (L1 and L3) that separate the VRs, allowing for proper folding of the molecule. The constant regions of the heavy chain, VR2 and the linker connecting them (L4) form the inner domain, while the outer domain is composed of the constant regions of the light chain, VR1 and their linker (L2).

[0078] While several modalities 210-240 are described herein, this is merely exemplary and it should be understood by one of skill in the art that the invention is not so limited and other types of antibody architectures / formats may be applied to the PK pipeline and corresponding PK models generated therefor depending on the application requirements.

[0079] 2a and 2b illustrate different types of modalities that may be used to generate a CODV-based dataset with known PK values ​​or PK characteristics for generating a PK model, and / or a CODV dataset with unknown PK values ​​or PK characteristics for evaluation and / or screening using a PK model to determine PK characteristics prior to in vitro and / or in vivo testing of the screened candidate CODV antibodies.

[0080] Figure 2c shows a multispecific CODV Ig-like dataset 250 that includes multiple CODV antibody molecules including 30 different CODV Ig-like antibodies from a diverse selection of target specificities and formats shown in CODV dataset table 251. In this example, the CODV dataset consists of 30 CODV antibodies. 23 CODV antibodies are bispecific (including two CODV formats, CODV-IgG and CODV-FcOL) and 7 CODV antibodies are trispecific (combining one CODV-Fab with one standard Fab arm). These three CODV modalities are shown in Figure 2b.

[0081] A CODV clearance table 251 and corresponding known PK clearance values ​​(mL / h / kg) (i.e., PK values) for 30 CODV molecules represented by CODV IDs 1-30. A CODV clearance distribution graph 252 shows the distribution of CODV clearance values ​​for CODV antibodies from the CODV clearance table 251. In this example, a PK clearance threshold 253 is set at 1 mL / h / kg, shown by the dashed line, which in this example can be used to define whether a CODV antibody is in the slow or fast clearance group of CODV 254 and 255.

[0082] That is, if the PK clearance value of the CODV antibody molecule is above the PK clearance threshold, the CODV antibody molecule is considered to be a fast clearance CODV antibody molecule, or part of fast clearance population 255. Similarly, if the PK clearance value of the CODV antibody molecule is equal to or below the PK clearance threshold, the CODV antibody molecule is considered to be a fast clearance CODV antibody molecule, or part of fast clearance population 254. Note that this PK clearance threshold may be set to any other value and / or may be set to define a population, multiple clearances, or used to label / characterize the PK clearance of different populations of CODV antibody molecules. Other PK clearance thresholds can be set depending on the type of antibody, for example, PK clearance thresholds in the region of 0.32 mL / h / kg are cited from the literature for mAbs (Avery et al., MAbs. 2018 Feb-Mar;10(2):244-255), but this typically depends on the protocol used to determine the PK clearance value and approaches may vary for different thresholds. The PK clearance thresholds can be set by the user of the system or can be defined by the PK value requirements of in vitro wet lab analysis and / or in vivo testing that the candidate CODV molecule must meet.

[0083] The CODV dataset 250 of Figure 2c is used herein as an exemplary CODV dataset to illustrate the pipeline process 100, the resulting PK model process 110, and the process 120 for use in selecting candidate PK models having unknown PK values ​​associated with the PK models. Although an exemplary CODV dataset of 20 CODV-like antibody molecules is described herein, this is by way of example only, and it should be understood by one of skill in the art that the invention is not so limited, and that any antibody dataset having multiple antibodies and known PK values ​​(e.g., clearance and / or other PK characteristics) can be used to generate a PK model as described herein and associated for use in predicting or estimating corresponding PK values ​​of an unknown antibody set and / or one or more unknown antibodies, etc.

[0084] FIG. 3a shows a system 300 for generating a PK model for predicting the PK value of an antibody. The system 300 includes a 3D modeling module 301, a 3D surface property module 302, a PK model generation module 303, and a PK model module 304 connected to each other. The functions of the 3D modeling module 301, the 3D surface property module 302, the PK model generation module 303, and the PK model module 304 may implement the steps and / or functions of the PK model pipeline process 100, the PK model process 110, and the antibody shortlisting process 120 of FIG. 1a, FIG. 1b, and / or FIG. 1c. In operation, the 3D modeling module 301 is configured to model the 3D structure of each antibody. For example, modeling the 3D structure of a set of CODV antibody molecules may be based on their amino acid sequences. The 3D surface property module 302 is configured to receive data representing the 3D structure of the antibodies and calculate area surface properties for the regions of interest for each antibody (e.g., steps 101, 102 and 103 of the PK model pipeline process 100). The PK model generation module 303 may be configured to generate a PK model based on steps 104 and 105 of the PK model pipeline process 100 of FIG. 1a. The PK model module 304 includes one or more generated PK models configured to receive the calculated area surface properties from the 3D surface property module 302 for the antibody of interest and apply them to one of the PK models to predict or estimate a PK value for the antibody of interest. The system 300 may further include an output module for displaying or transmitting data representing the predicted or estimated PK value of the antibody of interest to an operator or user of the system 300.

[0085] The PK model module 304 may further include functionality (e.g., one or more steps of process 120) for selecting a shortlist of candidate antibodies (e.g., CODV antibodies) from the plurality of antibodies based on desired PK features / characteristics. In operation, each of the candidate antibodies is input to the 3D modeling module 301 for modeling the 3D structure of each antibody of interest as an antibody of interest. The 3D models of the antibodies of interest are provided to the 3D surface property module 302 for calculating area surface properties of regions of interest for each antibody of interest. The PK model module 304 receives the calculated area surface properties from the 3D surface property module 302 for the antibody of interest and applies them to one of the PK models for predicting or estimating a PK value for the antibody of interest. The PK model module 304 may further include a PK comparison module (not shown) configured to compare the predicted or estimated PK value of the candidate antibody with the desired PK characteristics and select a candidate antibody from the plurality of candidate antibodies having a PK value that satisfies the one or more desired PK characteristics. The output module may be configured to output a shortlist of candidate antibodies from the selected candidate antibodies based on the comparison for use in in vitro wet lab analysis and / or in vivo testing.

[0086] FIG. 3b illustrates an in silico computational pipeline 310 for use in the system 300 for generating a PK model for predicting PK values ​​of an antibody of interest. The computational pipeline 310 is based on the PK model pipeline process 100 of FIG. 1a with further modifications of steps 101-105 of the PK model pipeline process 100. In this example, the computational pipeline 310 is described with reference to a CODV antibody and a basic CODV-Fab structure. The computational pipeline 310 is configured to generate a PK clearance model for predicting the clearance of a CODV antibody. Although PK clearance and PK clearance models are described herein, this is merely exemplary, the invention is not so limited, and it should be understood by one of skill in the art that the computational pipeline 310 and resulting PK model may be applied to any type of antibody and PK, PK value, or PK characteristics as the application requires. In this example, it is assumed that a PK clearance model is generated using multiple CODV antibodies with known PK clearance values. For example, the CODV antibody molecule dataset 250 having CODV IDs 1-20 described with reference to Figure 2c can be used as a plurality of CODV antibodies with known PK values ​​(e.g., PK clearance in this example). The computational pipeline 310 can include the following steps.

[0087] In step 311, the amino acid sequence of each CODV antibody from the CODV dataset with known PK values ​​is input to a 3D modeling module 301 where a 3D modeling system generates a 3D model of each CODV antibody. For example, the amino acid sequence may be input using FASTA format. Although a FASTA format is described herein, this is merely exemplary and the invention is not so limited, and it should be understood by those skilled in the art that any other suitable amino acid sequence data format may be used as required by the application. For example, the software used to generate each 3D model of each CODV may be, but is not limited to, Molecular Operating Environment (MOE), and / or any other suitable 3D modeling system or software as required by the application.

[0088] For the CODV-Fab structure, the workflow begins with the generation of a homology model of the CODV-Fab arm. Homology modeling can be performed using MOE. CDRs, FWs and linkers are identified from the sequence using the Chemical Computing Group (CCG) annotation numbering scheme. Although the CCG annotation numbering scheme is described herein, it is merely exemplary and the present invention is not so limited, and it is understood by those skilled in the art that any other suitable annotation numbering scheme can be used according to the requirements of the application. Among the five published crystal structures of CODV-Fab arms (e.g., PDB codes: 6O8D, 5FHX, 5WHZ, 6O89, and 5HCG), the one with the shortest difference in linker length is used as a template. After selection of the best CODV-Fab arm template, the inner and outer domains are modeled independently and assembled with the connecting linker. The best template from the MOE antibody database is used to model the VR. Finally, the entire structure is protonated at a defined pH (e.g., at physiological pH 7.4) and a series of minimization protocols (rigid body and free minimization) are applied to generate a high-quality 3D model using the AMBER10:EHT force field.

[0089] In step 312, the 3D modeling module 301 may be further configured to model the 3D structure of each CODV antibody based on one or more regions of interest (or a 3D modeling system). In this case, the portions or regions of interest of the CODV antibody being modeled are the CH1, VH2, VH1, CL, VL1 and VL2 domains and their connecting linkers (L1, L2, L3 and L4). Further regions of interest are shown in FIG. 3c.

[0090] After the 3D model of each CODV antibody is generated in step 313, the 3D model structure of each CODV antibody in the dataset is passed to the 3D surface properties module 302, where for each CODV antibody, the surface side chains (amino acids) of each 3D model of said each CODV antibody are sampled (using MOE) to generate an ensemble of structures aimed at accurately describing the patch surface map of said 3D model structure of said each CODV antibody. In this example, a number of samples N=50 is used such that 50 different sampled structures are modeled for each CODV antibody of multiple CODVs with known PK values ​​(e.g., CODV IDs 1-20 in FIG. 2c). While N=50 samples are used in this described example, it should be understood by those skilled in the art that this is merely an example and the invention is not so limited, and any suitable number of samples N>0 can be used.

[0091] The patch surface landscape may be influenced by the conformation of the CODV side chains, which may adopt multiple energetically favorable orientations. Therefore, using a single structural model to analyze the properties of a molecule's surface may not fully represent reality. Therefore, a set of N=50 different conformers for each CODV-Fab arm model of each CODV antibody in the dataset 250 was generated by using a sampling protocol at a defined pH (e.g., physiological pH 7.4). Once the conformers are generated, all surface patches are calculated. The surface properties of a patch may be hydrophobic, negatively or positively charged, depending on its physicochemical properties.

[0092] Each patch has an associated surface area (Å 2 The patch distribution is measured in Å or squared Å and has a list of participating residues. All patches belonging to each FW, CDR and linker (16 FWs, 12 CDRs, and 4 linkers for one CODV-Fab arm) were identified by annotating the CODV-Fab arm sequence using, for example but not limited to, the CCG annotation scheme or any other suitable annotation scheme. Then, the total hydrophobic, negatively charged, and positively charged regions from all FWs, CDRs, and linkers are obtained for all conformers. Finally, the average values ​​are calculated over all conformer structures to obtain the representative hydrophobic, ionic negatively charged, and ionic positively charged patch areas (Å) from all FWs, CDRs, and linkers for all conformations calculated for each CODV antibody in the dataset 250. 2 or squared angstroms), which is performed in steps 314-315 below.

[0093] In step 314, surface patches are calculated by the 3D surface property module 302 for all sampled structures of each of the CODV antibodies, the surface patches including positively charged patches, negatively charged patches, or hydrophobic patches. In other words, the surface patches include surface properties including ionic surface properties including ionic positive charges and ionic negative charges or hydrophobic surface properties.

[0094] In step 315, the 3D surface property module 302 calculates the surface patches (ion positive, ion negative, and hydrophobic), their surface areas (Å 2 The surface properties of each region of interest, e.g., framework (abbreviated as FW or FR), complementarity determining region (CDR), linker or combination of such regions, are determined by summing / adding the sizes of the surface properties of the patches which coincide with and / or overlap with each region of interest.

[0095] As an example, if the surface characteristic of at least one of the patches on the surface of each antibody is an ionic surface characteristic, the ionic surface characteristic of each patch includes a positively charged ionic region or a negatively charged ionic region. The ionic surface characteristic of each region of interest includes a positive ionic region surface characteristic and a negative ionic region surface characteristic. Calculating the region surface characteristic of each region of interest of each of the antibodies may further include calculating the positive ionic region surface characteristic of each region of interest of each of the antibodies by aggregating or averaging the ionic region surface characteristic of each region of interest of each of the antibodies with patches identified as being associated with (e.g., co-located and / or overlapping) each region of interest having a positive ionic region. Calculating the negative ionic region surface characteristic of each region of interest of each of the antibodies by aggregating or averaging the patches identified as being associated with (e.g., co-located and / or overlapping) each region of interest having a negative ionic region.

[0096] In a further example, the surface characteristic of at least one of the plurality of patches on the surface of each antibody further comprises a hydrophobic surface characteristic. The hydrophobic surface characteristic of each patch comprises an estimated region of hydrophobicity of said each patch. The at least one region surface characteristic for each region of interest further comprises a hydrophobic region surface characteristic. Calculating the hydrophobic region surface characteristic of each of the regions of interest by aggregating or averaging the hydrophobic surface characteristics of patches identified as associated with (e.g., co-located and / or overlapping with) said each region of interest.

[0097] In step 316, for each CODV antibody of the plurality of CODV antibodies, the surface properties of the regions of interest can be passed to the PK model generation module 303. Once the surface properties of the patches are mapped and combined with the surface properties for each of the regions of interest, multiple linear correlations for all combinations of the regions of interest between the surface properties of the regions of interest and the experimental PK values ​​(in this case, PK clearance) are performed for the plurality of antibodies in the CODV dataset. The multiple linear combinations also include calculating additional combinations of surface properties of the regions of interest for different groupings of the regions of interest based on sums / differences / ratios / fractions from combinations of two or more surface properties of different regions of interest.

[0098] For example, Figure 3c shows regions of interest in a CODV antibody structure 320. Regions of interest include, but are not limited to, linker 321 (L1, L2, L3, and L4), VR domains 322 including VR1 and VR2 domains, VH / VL subdomains 324 including VH1, VL1, VH2, VL2, and CDR / FW 326 for each of VH / VL domains 324, which include multiple regions 328 VH1-CDR1, VH1-CDR2 through VL2-FW3 and FW4-VL2 regions (e.g., there are 28 VH1 / VL1 / VH2 / VL2-CDR1-3 / FW1-4 regions). Referring to FIG. 3b, in step 316, the average hydrophobicity, ionic negative charge and ionic positive charge regions (e.g., from N=50 3D models) located in each region of interest (e.g., FW, CDR and linker) were used to calculate a descriptive score or scores based on the average properties of the CODV-Fab arm surface. In this regard, sums and ratios of many different sets of combinations of regions of interest (e.g., FW and CDR regions 326 and VH1 / VL1 / VH2 / VL2-CDR1-3 / FW1-4 regions 328) were performed to find the best grouping of regions of interest. That is, the best grouping of regions of interest for each antibody is the same grouping that results in the highest correlation (e.g., positive or negative correlation) between the descriptive score of that grouping and the experimental PK clearance values ​​of the dataset 250. For example, this may include the ratio of ionic patches (e.g., patches with ionic positive and / or ionic negative surface properties) at a particular location, or the total hydrophobic region from a particular region. After performing this task for all possible groupings or combinations of regions of interest, a score (e.g., average surface property) is generated that describes the subdomains (VH1, VL1, VH2, and VL2) and linkers of each VR and their local subregions (CDRs, FWs, or sets of linkers).

[0099] The in silico scores were used to perform a correlation with the experimental PK clearance dataset 250. For the CODV dataset 250, a strong exponential relationship was observed in the correlation, indicating that the experimental PK clearance dataset 250 should be transformed to its natural logarithm (Ln) counterpart to perform a linear correlation (positive 329a, negative 329b, or no correlation 329c). This may or may not necessarily apply to other datasets such as antibodies. In this example, the Pearson correlation coefficient (r), Spearman's rank correlation coefficient (r s ), and the coefficient of determination (r 2 ) were also calculated. While these types of correlation coefficients were calculated, this is merely exemplary and the invention is not so limited, however, one of skill in the art would understand that any suitable type of correlation coefficient may be applied and / or used. The grouping of the regions of interest that produced the score or combination score associated with the maximum correlation or maximum absolute correlation (e.g., maximum positive correlation 329a or maximum negative correlation 329b) is selected along with its corresponding linear correlation relationship to form a PK model. The relationship associated with the final combination score may be used by the PK model and may be used in part of the PK region surface property relationships of the PK model to predict or estimate PK clearance values ​​of unknown CODV antibodies associated with the types of CODV antibodies in the dataset 250.

[0100] In summary, for the CODV dataset 250 in Figure 2c, the PK properties of the CODV-Fab arms were profiled using an in silico pipeline 310. The pipeline 310 requires the sequence of the CODV-Fab arms to generate a homology model using MOE in steps 311-312. The relevant region names in the three-dimensional structure are indicated. Then, in step 313, an ensemble of structures (e.g., N=50) is generated by sampling the side chains of the model, and the surface patches of all conformers are quantified. Next, in steps 314 and 315, the total hydrophobic, positive and negative charge surface properties are calculated for each region of interest, such as FWs and CDRs in each domain (VH1, VL1, VH2, and VL2) and each linker and / or combination thereof (see table in step 315). The intensity of grey in the heatmap indicates the size of the total area of ​​that region of interest (darker is larger). In step 316, multiple linear correlations (or non-linear correlations) are performed for all combinations of same type surface properties of the regions of interest to find groupings from the regions of interest with the greatest correlations (positive or negative correlations 329a or 329b) for each type of surface property (e.g., ionic and hydrophobic surface properties) with experimental PK clearance values ​​across all CODV antibody datasets 250. This includes combining same type surface properties, such as positive and negative ionic ratios, by adding, subtracting, multiplying, and / or dividing in various suitable ways.

[0101] Alternatively or additionally, determining groupings from the regions of interest may include determining ion groupings by combining, for each grouping, the corresponding ion region surface properties (e.g., ion positive and ion negative) of each grouping for each of the plurality of antibodies. This may include performing the following steps: aggregating or averaging the positive ion region surface properties of the regions of interest in each grouping; aggregating or averaging the negative ion region surface properties of the regions of interest in each grouping. From this, determining a combined ion region surface property of each of the groupings based on calculating a ratio of the aggregated positive ion region surface properties of the groupings with the sum of the aggregated positive ion region surface properties of the groupings and the aggregated negative ion region surface properties of the groupings. Selecting, from all combinations of ionic groupings of the regions of interest, the ionic grouping that results in the greatest correlation (e.g., positive or negative correlation) between the determined combined ionic region surface properties of the groupings for each of the plurality of antibodies and the corresponding experimental PK clearance value (or any other PK value) for each of the plurality of antibodies.

[0102] Alternatively or additionally, determining the groupings from the regions of interest further comprises determining a hydrophobic grouping by combining, for each grouping, the corresponding hydrophobic region surface properties of each grouping for each of the plurality of antibodies based on an aggregation or averaging of the corresponding hydrophobic region surface properties in each of said groupings. From all combinations of hydrophobic groupings of the regions of interest, selecting the hydrophobic grouping that results in a maximum correlation (e.g., a positive or negative correlation) between the determined combined hydrophobic region surface properties of the groupings for each of the plurality of antibodies and the corresponding experimentally determined PK value for each of the plurality of antibodies.

[0103] The grouping of regions of interest found for the CODV dataset 250 with the greatest correlation with clearance included the VL1-CDR1 and CDR3 regions of interest for ionic surface properties, and the VR2-FW region of interest for hydrophobic surface properties. Two different types of surface properties were identified to be maximally correlated with PK clearance. These include ionic surface properties based on the positive ion ratio of the VL1-CDR1 & CDR3 regions (e.g., positive and negative ionic charges) and the hydrophobic area (Å) of the VR2-FW region. 2 ) were included as hydrophobic surface properties.

[0104] For the CODV dataset 250 in Figure 2c, all linear correlations of the in silico scores were analyzed, and two different types of surface properties (e.g., ionic and hydrophobic) yielded the greatest correlations for groupings of regions of interest for that type of surface property from all different combinations of regions of interest for that type of surface property. These included the positive ion ratios of VL1-CDR1 and CDR3 (Equation 1) and the hydrophobic area of ​​VR2-FW (Equation 2).

number

[0105] These relationships relating to the ion scores (e.g., Equation 1) and the combined hydrophobicity scores (e.g., Equation 2) can be used by the PK model and used as part of a PK region surface property relationship to predict or estimate PK clearance values ​​of unknown CODV antibodies associated with the CODV antibody types in dataset 250.

[0106] 4a depicts a set of plots 400 of correlations between experimental PK clearance values ​​and ionic properties in two regions of interest in the CODV Ig-like molecule dataset 250, namely CDR1 and CDR3 of the VL1 domain. Plot 402a depicts the positive ionic charge region (Å) of the VL1-CDR1 region. 2, or measured in squared angstroms) and the experimental PK clearance values ​​(in vivo clearance ln (mL / h / kg)) of the antibodies in dataset 250. Plot 402b shows the correlation between the negative ionic charge area (Å) of the VL1-CDR1 region. 2 , or measured in squared angstroms) and the experimental PK clearance values ​​(in vivo clearance ln(mL / h / kg)) of the antibodies in dataset 250. Plot 404a shows the correlation between the positive ionic charge area (Å) of the VL1-CDR3 region. 2 , or measured in squared angstroms) and the experimental PK clearance values ​​(in vivo clearance ln(mL / h / kg)) of the antibodies in dataset 250. Plot 402b shows the correlation between the negative ionic charge area (Å 2 , or measured in squared angstroms) and the experimental PK clearance values ​​(in vivo clearance ln (mL / h / kg)) of the antibodies in dataset 250. As can be seen in plots 402a and 404a, the slow PK clearance molecules have smaller positive areas (Å) in CDR1 and CDR3 of the VL1 domain. 2 ), whereas fast PK clearance molecules behave differently. 2 Completely opposite behavior was observed for plots 402b and 404b with r, ... s , and r 2Coefficients are displayed in all cases. The best score in terms of correlation coefficient is the positive ion ratio in the VL1-CDR1 and CDR3 grouping, and therefore this relationship can be selected for use in forming a PK region surface property relationship for use in a PK model to predict or estimate the PK clearance of CODV with unknown PK clearance values.

[0107] 4b depicts a set of plots 410 of correlations between experimental PK clearance values ​​and hydrophobicity in regions of interest, i.e., FW in VH2, VL2, and VR2, in the CODV Ig-like molecule dataset 250. Plot 412a depicts the hydrophobic area (Å) of the VH2-FW region. 2 , or measured in squared angstroms) and the experimental PK clearance values ​​(in vivo clearance ln(mL / h / kg)) of the antibodies in dataset 250. Plot 412b shows the correlation between the hydrophobic area (Å 2 , or measured in squared angstroms) and the experimental PK clearance values ​​(in vivo clearance ln(mL / h / kg)) of the antibodies in dataset 250. Plot 412c shows the correlation between the hydrophobic area (Å) of the VR2-FW region. 2 , or measured in squared Angstroms) and the experimental PK clearance values ​​(in vivo clearance ln(mL / h / kg)) of the antibodies in dataset 250. These hydrophobicity scores for this region of interest can be combined to obtain the maximum correlation obtained according to Equation 2: r, r s , and r 2 Coefficients are displayed in all cases. Slow PK clearance molecules have smaller hydrophobic regions (Å 2 ), whereas fast PK clearance molecules behave differently. 2 ) showed the highest correlation coefficient out of the three hydrophobicity scores.

[0108] Figure 4c shows a cluster plot 420 in which the two best in silico scores from plot 406 of Figure 4a and plot 412c of Figure 4b are plotted against each other. That is, plot 420 shows the ratio of positive ions in VL1-CDR1 and CDR3 (e.g., plot 406 of Figure 4a) versus the hydrophobic region (Å) in VR2-FW (e.g., plot 412c of Figure 4b). 2 ). The marker shapes indicate slow (e.g., circular) and fast (e.g., triangular) PK clearance molecules. The size of the marker represents the magnitude of the PK clearance value for that antibody (the larger the marker size, the faster the PK clearance). As can be seen, the slow PK clearance molecules tend to cluster together in cluster group 420b in the lower left corner of plot 420, and the fast PK clearance molecules tend to cluster together in cluster group 420a in the upper right region of plot 420.

[0109] These two selected in silico scores represent different regions (VR1 and VR2) and physicochemical properties (ionicity and hydrophobicity) of the CODV-Fab arm that correlate with PK clearance for the data set 250. After plotting the positive ion ratios of VL1-CDR1 and CDR3 against the hydrophobic region of VR2-FW in plot 420, two clusters 420a and 420b enriched for slow and fast clearance CODV Ig-like molecules were identified. Both in silico scores are used to calculate a combined score (referred to as "in silico clearance likelihood"). The terms "combined score" and "in silico clearance likelihood" are used interchangeably throughout this disclosure. For the construction of the combined score (in silico clearance likelihood) obtained according to Equation 3, the two in silico descriptors were weighted equally.

number

[0110] The in silico score, ion ratio, on the one hand, is a ratio (a value between 0 and 1), and the combined hydrophobic averaged properties on the other hand are of a larger value (approximately 500 Å 2~about 2500Å 2 Assuming that the mean ionic and hydrophobic scores are in the range of 0 to 1, the scaled percentile of the latter is calculated, which ranges between 0 and 1. This facilitates obtaining a combined score that represents both the ionic and hydrophobic scores at once with equal importance (50% each). The correlation between the combined scores and the experimental PK clearance values ​​for dataset 250 is shown in Figure 4d.

[0111] Figure 4d shows the experimental PK clearance values ​​of dataset 250 combined with the two best in silico scores: the positive ion ratios of VL1-CDR1 and CDR3 and the hydrophobic area (Å) of VR2-FW. 2 4 is a plot 430 showing the correlation between the combination scores obtained from r, r s , and r 2 The coefficients are displayed. Slow PK clearance molecules showed lower combined scores (in silico clearance likelihood) than fast PK clearance molecules. The relationship associated with the combination of ionicity and hydrophobicity scores (e.g., Equation 3) can be used by a PK model as a PK region surface property relationship to predict or estimate the PK clearance value of an unknown CODV antibody associated with a type of CODV antibody in the dataset 250. Thus, according to the processes 100, 110, 120 of Figures 1a-1b and the system 300 and pipeline 310 of Figures 3a and 3b, a PK model can be generated based on the combined ionicity and hydrophobicity scores (e.g., Equation 3) and used to predict or estimate the PK clearance value of an unknown CODV antibody as described with reference to the processes 110 and 120 of Figures 1b and 1c. PK models may be used for technical applications such as, but not limited to, evaluation of antibodies with unknown PK values ​​(e.g., clearance) for short-listing prior to in vitro wet lab analysis and / or in vivo testing.

[0112] FIG. 5 shows the hydrophobic regions (Å) in the VR2-FW values ​​of two different test CODV molecules not within the data set 250. 25 is a plot 500 of the positive ion ratios in VL1-CDR1 and CDR3 versus VL1-CDR2. In this case, two different test CODV molecules have the same binding domains, but in different orientations. The first test CODV molecule (CODV-AB, 502a) is a CODV containing V domains A and B. The second test CODV molecule (CODV-BA, 502b) is a CODV containing the same V domains A and B, but in a different orientation compared to CODV-AB. These were used as test cases because V domain A is known to have better PK properties (e.g., better PK clearance) than V domain B. Additionally, it is also known that V domain B has a larger positive surface area on its surface (including the CDRs) than V domain A. As can be seen, CODV-AB with a "good" V-domain A as the VL1 domain (e.g., CODV 502a) are CODV-AB that exhibit better in silico surface property parameters, including lower positive ion ratios for VL1-CDR1 and CDR3, and lower hydrophobicity for VR2-FW.

[0113] In contrast, CODV-BA (e.g., CODV 502b) with a "bad" V domain B as the VL1 domain has a very large positive ion ratio in VL1-CDR1 and CDR3, and a larger hydrophobic region in VR2-FW. Furthermore, in vitro assays were performed on these two CODV antibodies 502a and 502b to predict in vivo clearance based on FcRn chromatography data. FcRn chromatography data represents retention time, which correlates with PK clearance. Considering this, the longer the retention time in FcRn data, the worse the expected clearance pattern or PK clearance value of the antibody. When analyzing the in vitro FcRn chromatography data of these two CODVs, it was found that the expected "good" CODV-AB antibody 502a has a shorter FcRn retention time than the CODV-BA antibody 502b, which is consistent with the PK model of the CODV molecule.

[0114] These results indicate that the PK model should correlate with other future unknown CODVs with unknown clearance values. Thus, there is an alignment between the in silico PK model, the in vitro FcRn wet lab analysis, and the in vivo PK clearance properties. Although the in vivo experimental PK clearance values ​​for these CODV antibodies 502a and 502b are not known, the PK model predicts that CODV 502a will have a better PK clearance value than CODV 502b, which is also shown by FcRn chromatography. This indicates that the PK model as constructed or formed according to the PK model pipeline processes 100 and 300 of Figures 1a and 3a and applied according to processes 110, 120 and described with reference to Figures 3a-4d can make accurate predictions of the PK properties of unknown CODV antibodies for shortlisting for in vitro wet lab analysis and / or in vivo testing.

[0115] The concept shown in Figure 5 was experimentally verified by the data shown in Figure 6. Figure 6 shows two plots: plot 602a and plot 603a. Plot 602a shows the hydrophobic region (Å) at VR2-FW values ​​for two different CODV molecules, each containing V domains C and D, but with different orientations. 2 ) of VL1-CDR1 and CDR3. Two different orientations of two CODV molecules are CODV-CD (602b) and CODV-DC (602c). From plot 602a, it can be seen that the CODV-DC (602c) antibody has a higher positive ion ratio in VL1-CDR1 and CDR3 and a higher hydrophobic region (Å) in VR2-FW. 2 ), and therefore is predicted to have a better PK clearance value than its counterpart CODV-CD(602b).

[0116] Plot 603a shows the combined score (likelihood of in silico clearance) versus experimental clearance (in vivo clearance ln (mL / h / kg)) for the same two different CODV molecules as plot 602a. Plot 603a shows that the combined score (likelihood of in silico clearance) for CODV-DC (603c) is lower than that for CODV-CD (603b) and correlates with the experimental clearance (in vivo clearance ln (mL / h / kg)). CODV-DC (603c) exhibits slower clearance than CODV-CD (603b).

[0117] FIG. 7 includes a dataset 700 and a plot 701. The dataset 700 represents multiple multispecific CODV Ig-like antibodies, including seven different antibodies not represented in the dataset 250 from a diverse selection of target specificities. The CODV-IgG modality is shown in the architecture / structure 220 of FIG. 2b. The plot 701 shows the correlation between in vivo clearance ln(mL / h / kg) and the combination score (in silico combination likelihood) obtained from Equation 3. The threshold 702 indicates molecules predicted to exhibit a slow clearance profile (<0.5 units in the in silico clearance likelihood) or a fast clearance profile (less than 0.5 units in the in silico clearance likelihood). The threshold 703 indicates molecules with slow experimental clearance (less than 1 mL / h / kg, or less than 0 as its natural logarithmic counterpart) and molecules with fast experimental clearance (more than 1 mL / h / kg, or more than 0 as its natural logarithmic counterpart). r,r s , and r 2 The coefficients are displayed.

[0118] The predicted PK values ​​of the molecules incorporated into dataset 700 and shown in plot 701 were generated according to PK model pipeline processes 100 and 300 of Figures 1a and 3a and applied according to processes 110, 120 as described with reference to Figures 3a-4d. At the time of their prediction, no PK data (e.g., half-life, clearance) was available for any of the seven molecules incorporated into dataset 700. These molecules were purposefully selected from a larger panel of available CODV-IgGs based solely on the PK model pipeline of the present invention, in order to validate the method with molecules with unknown PK data. This effort was aimed at identifying four molecules with a slow clearance profile and three molecules with a fast clearance profile. To experimentally determine in vivo clearance, seven selected CODV-IgG antibodies were expressed in HEK293 cells, purified in two steps, and characterized for PK clearance in human FcRn transgenic mice (Tg32 hFcRn SCID strain) according to the Animal Use Protocol (AUP) and Institutional Animal Care and Use Committee (IACUC) regulations.

[0119] Plot 701 shows a very strong correlation between experimental clearance (in vivo clearance ln(mL / h / kg)) and the combined score (in silico clearance likelihood). All seven molecules represented in dataset 700 and plotted in plot 701 were predicted by the PK model pipeline processes 100 and 300 of Figures 1a and 3a, applied according to processes 110, 120, and behaved as expected as described with reference to Figures 3a-4d. The PK models and / or processes described herein provide the advantage of an efficient and accurate mechanism to substantially reduce the number of candidate antibodies and / or CODV antibodies for shortlisting them for wet lab analysis. This then allows for a large number of antibodies to be evaluated and screened, and only those antibodies with the required estimated / predicted PK values / characteristics (e.g., PK clearance) can be shortlisted prior to expensive and / or cumbersome in vitro wet lab analysis and / or in vivo testing.

[0120] While the PK models and processes 100, 110, 120 and 310 and system 300 are described with reference to antibodies or CODV antibodies / CODV dataset 250 / CODV dataset 700, etc., this is by way of example only and the invention is not so limited, and it should be understood by one of skill in the art that the PK models and processes 100, 110, 120 and 310 and system 300 described herein may be applicable to any molecule that may be described in terms of an amino acid sequence, such as, but not limited to, antibodies, CODV antibodies, enzymes, other proteins, and / or any other suitable protein format, etc. While PK values ​​are described herein in connection with the generation of the PK models, this is by way of example only and the invention is not so limited, and it should be understood by one of skill in the art that any other property of interest other than PK values ​​may be used, such as, but not limited to, oligomerization, protein expression, etc., and / or as required by the application.

[0121] 8 is a schematic diagram of a system / apparatus for performing the methods described herein. The illustrated system / apparatus is an example of a computing device. Those skilled in the art will appreciate that other types of computing devices / systems, such as distributed computing systems, may alternatively be used to perform the methods described herein.

[0122] The device (or system) 800 includes one or more processors 802. The one or more processors control the operation of other components of the system / device 800. The one or more processors 802 may include, for example, a general purpose processor. The one or more processors 802 may be single-core or multi-core devices. The one or more processors 802 may include a central processing unit (CPU) or a graphical processing unit (GPU). Alternatively, the one or more processors 802 may include specialized processing hardware, such as a RISC processor or programmable hardware with embedded firmware. Multiple processors may be included.

[0123] The system / apparatus includes a working or volatile memory 804. One or more processors may access the volatile memory 804 to process data and may control data storage in the memory. The volatile memory 804 may include any type of RAM, such as static RAM (SRAM), dynamic RAM (DRAM), or it may include a flash memory, such as an SD-card.

[0124] The system / apparatus includes a non-volatile memory 806. The non-volatile memory 806 stores a set of operating instructions 808 in the form of computer readable instructions for controlling the operation of the processor 802. The non-volatile memory 806 can be any type of memory, such as a read only memory (ROM), flash memory, or magnetic drive memory.

[0125] The one or more processors 802 are configured to execute operational instructions 808 to cause the system / device to perform any of the methods or processes described herein with reference to Figures 1a-7. The operational instructions 808 may include code for hardware components of the system / device 800 (i.e., drivers), as well as code for basic operation of the system / device 800. Typically, the one or more processors 802 execute one or more instructions of the operational instructions 808, which are stored permanently or semi-permanently in non-volatile memory 806, and the volatile memory 804 is used to temporarily store data generated during execution of said operational instructions 808.

[0126] Implementations of the methods described herein may be realized in digital electronic circuitry, integrated circuitry, specially designed ASICs (application specific integrated circuits), computer hardware, firmware, software, and / or combinations thereof, which may include a computer program product (e.g., software stored on a magnetic disk, optical disk, memory, programmable logic device, etc.) containing computer readable instructions that, when executed by a computer, cause the computer to perform one or more of the methods described herein, such as those described with respect to FIG.

[0127] System features described herein may also be provided as method features, and vice versa. As used herein, "means-plus-function" features may be expressed in terms of their corresponding structure. In particular, method aspects may also apply to system aspects, and vice versa.

[0128] Furthermore, any, some, and / or all of the features in one embodiment may be applied to any, some, and / or all of the features in any other embodiment, in any appropriate combination. It should also be understood that specific combinations of the various features described and defined in any embodiment of the invention may be implemented and / or provided and / or used separately.

[0129] While several embodiments have been shown and described, those skilled in the art will understand that changes could be made to those embodiments without departing from the principles of the present disclosure, the scope of which is defined in the claims.

Claims

1. A computer-based method for generating a pharmacokinetic model PK for predicting the PK value of a target antibody, Receiving an input dataset for multiple antibodies, wherein the input dataset includes data representing the amino acid sequence of each antibody and the experimentally determined PK value of each antibody; Calculating one or more surface properties for each of the antibodies based on the corresponding amino acid sequence; Based on the one or more surface properties calculated for each of the antibodies, one or more regional surface properties are calculated for one or more regions of interest for each of the antibodies; Determining a grouping from one or more regions of interest that produces the maximum correlation between the calculated region surface properties of the grouping and the corresponding experimentally determined PK values, thereby establishing a PK region surface property relationship for the plurality of antibodies; and A computer-aided method comprising generating a PK model for predicting the PK value for the antibody of the objective, wherein the PK model is configured to receive one or more input region surface characteristics of the determined groupings for the antibody of the objective, and to output a predicted PK value for the antibody of the objective by applying the input region surface characteristics to the PK region surface characteristic relationship.

2. The aforementioned PK model, One or more surface properties of the aforementioned antibody are calculated based on its amino acid sequence; The computer-aided method according to claim 1, further configured to calculate one or more regional surface characteristics for the antibody of interest for the determined grouping of the region of interest, based on one or more surface characteristics calculated for the antibody of interest, thereby providing the input regional surface characteristics.

3. The computer-aided method according to claim 1, wherein the generated PK model is further configured to predict a shortlist of candidate antibodies having desired PK properties for use in in vitro wet laboratory analysis and / or in vivo testing.

4. For each of the candidate antibodies, the method is as follows: A step of calculating one or more surface properties for each of the candidate antibodies based on the corresponding amino acid sequence; For each of the candidate antibodies, the step of calculating one or more regional surface properties for the determined grouping from the region of interest; In order to predict the PK value of each of the candidate antibodies, the data representing the calculated regional surface characteristics of each of the candidate antibodies is input into the generated PK model; A step of receiving the predicted PK value for each of the candidate antibodies as output from the PK model; and If the received predicted PK value of the candidate antibody exhibits one or more of the desired PK characteristics defined for the shortlist, the step of adding the candidate antibody to the shortlist of candidate antibodies; and Step 1: Output data representing a shortlist of candidate antibodies for use in at least in vitro wet laboratory analysis or in vivo testing. The computer implementation method according to claim 3, further comprising:

5. The computer implementation method according to claim 1, wherein calculating one or more surface properties for each of the plurality of antibodies includes modeling the three-dimensional molecular structure of each of the antibodies and calculating the distribution of one or more surface properties across the surface of the modeled molecular structure of each of the antibodies.

6. The computer-aided method according to claim 1, wherein the one or more calculated surface properties include one or more of the positively charged surface area, the negatively charged surface area, and the hydrophobic surface area.

7. The computer-aided method according to claim 5, wherein the surface of each of the modeled molecular structures of the antibody comprises a plurality of patches of one or more surface properties, each patch having a patch region based on the distribution of the surface properties in the modeled molecular structure, and the number of patches of each antibody is the same or different for each of the plurality of antibodies.

8. The computer implementation method according to claim 7, wherein the calculation of one or more regional surface properties for one or more regions of interest for each of the antibodies is based on the area of ​​the patch of one or more surface properties associated with each region of interest.

9. The computer-aided method according to claim 1, wherein the plurality of antibodies and the target antibody are crossover bivariate (CODV) antibodies.

10. The computer-aided method according to claim 9, wherein at least one region of interest is selected from complementary domain regions of a variable heavy chain VH domain or a variable light chain VL domain, a CDR, a framework region, or a linker, comprising CDR1, CDR2, CDR3, FW1, FW2, FW3, FW4 of either the VH domain or VL domain of two variable V domains of the CODV antibody.

11. The computer implementation method according to claim 10, wherein the grouping from the region of interest is CDR1 of VL1, CDR3 of VL1, and FW1-4 of VL2 and VH2 of the CODV antibody.

12. The computer implementation method according to claim 1, wherein the PK surface characteristic relationship is a linear PK surface characteristic relationship.

13. From one or more of the above regions of interest, the corresponding calculated region surface properties of the grouping and Determining the grouping that produces the maximum correlation with the corresponding PK value determined experimentally is The computer-aided method according to claim 1, comprising calculating a combination score from the corresponding calculated region surface characteristics of the grouping, determining the maximum correlation between the combination score and the experimentally determined corresponding PK value, and thereby establishing the PK region surface characteristic relationship of the plurality of antibodies.

14. The computer implementation method according to claim 13, wherein the combination score includes two equally weighted calculated regional surface characteristics.

15. An apparatus comprising a processor, a memory unit, and a communication interface, wherein the processor is connected to the memory unit and the communication interface, and the processor and memory are configured to perform the computer implementation method described in claim 1.

16. A computer-readable medium containing data or instruction code, which, when executed on a processor, causes the processor to perform the computer implementation method described in claim 1.

17. It is a system, A three-dimensional surface properties module configured to receive data representing multiple candidate antibodies and calculate the regional surface properties of each of the candidate antibodies' regions of interest; A pharmacokinetic PK model module configured to receive the calculated regional surface characteristics corresponding to each of the candidate antibodies for predicting the PK value of each of the candidate antibodies; A PK comparison module configured to compare the predicted PK value of the candidate antibody with a desired PK characteristic and select a candidate antibody from the plurality of candidate antibodies having a PK value that satisfies one or more of the desired PK characteristics; and A system including an output module configured to output a shortlist of candidate antibodies from the selected candidate antibodies based on the comparison for use in in vitro wet laboratory analysis and / or in vivo testing.

Images