Hybrid protein design

Abstract

A method may include applying a protein sequence calculation model to generate a plurality of proposed protein sequences based on an input protein sequence. A set of possible amino acid residues for each position in at least a portion of the output protein sequence may be identified based on the plurality of proposed protein sequences. A first protein structure having the output protein sequence may be generated by applying a protein structure calculation model to select an amino acid residue from a corresponding set of possible amino acid residues for each location in at least the portion of the output protein sequence. The protein structure calculation model may further determine a conformation of the amino acid residues selected for inclusion in the output protein sequence. In some cases, the first protein structure may be grafted onto a second protein structure to form a third protein structure.

Description

[0001] CROSS-REFERENCE TO RELATED APPLICATIONS

[0002] This application claims priority to U.S. Provisional Application No. 63 / 359,167, entitled “Hybrid Protein Design” and filed on July 7, 2022, the disclosure of which is incorporated herein by reference in its entirety. Technical Field

[0003] The subject matter described herein relates generally to protein design, and more particularly to protein design techniques that integrate protein sequence design with protein structure design. Background Art

[0004] Proteins are responsible for many important cellular functions, including, for example, enzymatic reactions, molecular transport, regulation and execution of many biological pathways, cell growth, proliferation, nutrient uptake, morphology, movement, intercellular communication, etc. A protein molecule may comprise one or more polypeptide chains, each of which comprises a plurality of amino acid residues linked together by peptide bonds. The primary structure of a molecule (which depends on the sequence of amino acid residues in the polypeptide chains that form the protein molecule) may refer to the three-dimensional structure (or conformation) of the protein molecule. For example, the secondary structure of a protein molecule may be defined, at least in part, by the torsion angles (or dihedral angles) of the peptide bonds present in the backbone (or main chain) of the protein molecule, while the tertiary structure of a protein molecule may be defined by the folding of the polypeptide chains.

[0005] The function of a protein molecule may depend on the sequence of amino acids in the polypeptide chain that forms the protein molecule and the three-dimensional structure formed by the polypeptide chain. Therefore, a key goal of protein design is to construct novel amino acid residue sequences (e.g., antibodies, etc.) that exhibit certain desired properties, including the ability to adopt a certain three-dimensional structure. In the case of macromolecular drug discovery, novel protein sequences can be designed to fold into a three-dimensional structure that is complementary to the three-dimensional structure of a target molecule (e.g., an antigen, such as a viral antigen, a tumor antigen, etc.) and is sufficiently stable to allow the corresponding protein molecule to bind to the target molecule. Summary of the invention

[0006] Systems, methods, and articles of manufacture, including computer program products, for hybrid protein design are provided. In some exemplary embodiments, a system including at least one processor and at least one memory is provided. The at least one memory may include program code that, when executed by the at least one processor, provides operations. The operations may include: identifying a protein sequence computational model and a protein structure computational model; applying the protein sequence computational model to generate a plurality of proposed protein sequences based at least on an input protein sequence; identifying a set of possible amino acid residues for each position in at least a portion of an output protein sequence based at least on the plurality of proposed protein sequences; and using the protein structure computational model to generate a first protein structure having an output protein sequence by: applying the protein structure computational model to select a possible amino acid residue from the set of possible amino acid residues for inclusion in the output protein sequence for each position in at least the portion of the output protein sequence.

[0007] In another aspect, a method for hybrid protein design is provided. The method may include: identifying a protein sequence computational model and a protein structure computational model; applying the protein sequence computational model to generate a plurality of proposed protein sequences based at least on an input protein sequence; identifying a set of possible amino acid residues for each position in at least a portion of an output protein sequence based at least on the plurality of proposed protein sequences; and using the protein structure computational model to generate a first protein structure having the output protein sequence by: applying the protein structure computational model to select a possible amino acid residue from the set of possible amino acid residues for each position in at least the portion of the output protein sequence for inclusion in the output protein sequence.

[0008] In another aspect, a computer program product is provided that includes a non-transitory computer-readable medium storing instructions. The instructions may cause operations that may be performed by at least one data processor. The operations may include: identifying a protein sequence computational model and a protein structure computational model; applying the protein sequence computational model to generate a plurality of proposed protein sequences based at least on an input protein sequence; identifying a set of possible amino acid residues for each position in at least a portion of an output protein sequence based at least on the plurality of proposed protein sequences; and using the protein structure computational model to generate a first protein structure having the output protein sequence by: applying the protein structure computational model to select a possible amino acid residue from the set of possible amino acid residues for inclusion in the output protein sequence for each position in at least the portion of the output protein sequence.

[0009] In some variations, one or more features disclosed herein, including the following features, can optionally be included in any feasible combination.

[0010] In some variations, the plurality of proposed protein sequences may be compared to generate an aligned plurality of protein sequences. The set of possible amino acid residues for each position in at least the portion of the output protein sequence may be identified based at least on the aligned plurality of protein sequences.

[0011] In some variations, the plurality of proposed protein sequences may be aligned by applying one or more of: dynamic programming, progressive alignment, hierarchical alignment, iterative alignment, motif discovery, deep learning models, and hidden Markov models.

[0012] In some variations, identifying the set of possible amino acid residues for each position in at least the portion of the output protein sequence may include identifying a first amino acid residue but not a second amino acid residue for inclusion in the set of possible amino acid residues.

[0013] In some variations, identifying the set of possible amino acid residues for each position in at least the portion of the output protein sequence may further include: determining a first frequency at which a first amino acid residue occurs at the position across the plurality of proposed protein sequences generated by the protein sequence computational model, determining a second frequency at which a second amino acid residue occurs at the position across the plurality of proposed protein sequences generated by the protein sequence computational model, and identifying the first amino acid residue but not the second amino acid residue for inclusion in the set of possible amino acid residues for the position based at least on the first frequency and the second frequency.

[0014] In some variations, a first amino acid residue can be identified for inclusion in the set of amino acid residues based at least on a first frequency satisfying one or more thresholds. A second amino acid residue can be identified for exclusion from the set of possible amino acid residues based at least on a second frequency failing to satisfy the one or more thresholds.

[0015] In some variations, identifying the set of possible amino acid residues for the position in the output protein sequence may further include determining the one or more thresholds based on at least one of: a maximum, a minimum, a median, a mean, and a mode of frequencies with which each amino acid residue in the plurality of amino acid residues appears at the position across the plurality of proposed protein sequences generated by the protein sequence computational model.

[0016] In some variations, the set of possible amino acid residues may include some, but not all, of the following: alanine, arginine, asparagine, aspartic acid, cysteine, glutamic acid, glutamine, glycine, histidine, isoleucine, leucine, lysine, methionine, phenylalanine, proline, serine, threonine, tryptophan, tyrosine, valine, selenocysteine, and pyrrolysine.

[0017] In some variations, the protein structure computational model may generate the first protein structure by at least determining the identity and conformation of the amino acid residue occupying each position in at least the portion of the output protein sequence based at least on the energy of the first protein structure having the output protein sequence.

[0018] In some variations, the protein structure computational model may determine the identity and conformation of the amino acid residue occupying each position in at least the portion of the output protein sequence by at least modifying at least one of the identity and the conformation of the amino acid residue to minimize the energy of the first protein structure.

[0019] In some variations, the computational model of the protein structure may modify at least one of the identity and conformation of the amino acid residue occupying the position by at least: (i) changing the conformation of the amino acid residue occupying the position, or (ii) selecting a different possible amino acid residue for the position from a set of amino acid residues associated with the position.

[0020] In some variations, the protein structure computational model may determine the identity and conformation of the amino acid residue occupying each position in at least the portion of the output protein sequence by at least: determining a first energy of a first protein structure having a first possible amino acid residue from the set of possible amino acid residues, determining a second energy of the first protein structure having a second possible amino acid residue from the set of possible amino acid residues, and generating the first protein structure to include the first possible amino acid residue instead of the second possible amino acid residue based at least on the first energy being lower than the second energy.

[0021] In some variations, the protein structure computational model may further determine the identity and conformation of the amino acid residue occupying each position in at least the portion of the output protein sequence by at least determining a third energy of the first protein structure having a first conformation of the first possible amino acid residue, determining a fourth energy of the first protein structure having a second conformation of the first possible amino acid residue, and generating the first protein structure to include the first conformation of the first possible amino acid residue instead of the second conformation of the first possible amino acid residue based at least on the third energy being lower than the fourth energy.

[0022] In some variations, a property analysis model may be applied to determine a property of each protein sequence included in the plurality of proposed protein sequences. At least one protein sequence in the plurality of proposed protein sequences may be identified for exclusion based at least on the property of each protein sequence. The at least one protein sequence may be excluded from the plurality of proposed protein sequences, and then the set of possible amino acid residues for each position in the output protein sequence may be identified based at least on the remaining plurality of proposed protein sequences.

[0023] In some variations, a first portion of a second protein structure may be identified. A third protein structure may be generated by replacing the first portion of the second protein structure with at least a second portion of the first protein structure.

[0024] In some variations, a third protein sequence generated as a third protein structure comprising a second portion of the first protein structure and a third portion of the second protein structure may be determined. A protein structure computational model and / or a different protein structure computational model may be applied to determine a fourth protein structure having at least the third protein sequence based at least on the third protein sequence. A similarity metric quantifying differences between the third protein structure and the fourth protein structure may be determined. The third protein sequence may be identified as a candidate for synthesis based at least on the similarity metric satisfying one or more thresholds.

[0025] In some variations, the second protein structure may be selected based at least on the second protein structure exhibiting one or more desired properties.

[0026] In some variations, the first portion of the second protein structure may comprise a first antigen binding site of a first antibody having the second protein structure, and the second portion of the first protein structure may comprise a second antigen binding site of a second antibody having the first protein structure.

[0027] In some variations, the first portion of the second protein structure may comprise a first paratope of the first antibody having the second protein structure, and the second portion of the first protein structure may comprise a second paratope of the second antibody having the first protein structure.

[0028] In some variations, the first portion of the second protein structure may comprise a first complementarity determining region (CDR) of the first antibody having the second protein structure, and the second portion of the first protein structure may comprise a second complementarity determining region (CDR) of the second antibody having the first protein structure.

[0029] Specific implementations of the current subject matter may include, but are not limited to, methods consistent with the description provided herein and articles including tangibly embodied machine-readable media that are operable to cause one or more machines (e.g., computers, etc.) to cause operations that implement one or more of the described features. Similarly, a computer system that may include one or more processors and one or more memories coupled to the one or more processors is also described. A memory that may include a non-transitory computer-readable or machine-readable storage medium may include, encode, store, etc., one or more programs that cause one or more processors to perform one or more of the operations described herein. A computer-implemented method consistent with one or more implementations of the current subject matter may be implemented by one or more data processors present in a single computing system or multiple computing systems. Such multiple computing systems may be connected and may exchange data and / or commands or other instructions, etc., via one or more connections, including, for example, via a network (e.g., the Internet, a wireless wide area network, a local area network, a wide area network, a wired network, etc.) via a direct connection between one or more computing systems in the multiple computing systems, etc.

[0030] Details of one or more variations of the subject matter described herein are set forth in the accompanying drawings and the following description. Other features and advantages of the subject matter described herein will become apparent with reference to the description and drawings, as well as the claims. Although certain features of the presently disclosed subject matter are described for illustrative purposes related to various protein design techniques that integrate protein sequence design with protein structure design, it should be readily understood that such features are not intended to be limiting. The claims that follow this disclosure are intended to define the scope of the protected subject matter. BRIEF DESCRIPTION OF THE DRAWINGS

[0031] The accompanying drawings, which are incorporated into and constitute a part of this specification, illustrate certain aspects of the subject matter disclosed herein and, together with the description, help explain some principles associated with the disclosed embodiments.

[0032] FIG1 depicts a system diagram showing an example of a protein design system according to some exemplary embodiments;

[0033] FIG2 depicts a flow chart showing an example of a process for hybrid protein design according to some exemplary embodiments;

[0034] FIG3A depicts a flow chart showing another example of a process for hybrid protein design according to some exemplary embodiments;

[0035] FIG3B depicts a flow chart showing another example of a process for hybrid protein design according to some exemplary embodiments;

[0036] FIG4A depicts a flow chart showing another example of a process for hybrid protein design according to some exemplary embodiments;

[0037] FIG4B depicts a schematic diagram showing an example of a process for validating a protein structure generated in silico, according to some exemplary embodiments;

[0038] FIG5 depicts a flow chart showing another example of a process for hybrid protein design according to some exemplary embodiments;

[0039] FIG6A depicts a schematic diagram showing examples of aligned protein sequences according to some exemplary embodiments;

[0040] FIG6B depicts a schematic diagram showing examples of possible amino acid residue groups for different positions in a protein sequence according to some exemplary embodiments;

[0041] FIG7A depicts a schematic diagram showing an example of a process for hybrid protein design, in which a protein sequence output by a protein sequence computational model undergoes structural design to generate a corresponding protein structure, according to some exemplary embodiments;

[0042] FIG7B depicts a schematic diagram showing another example of a process for validating a protein structure generated in silico, according to some exemplary embodiments;

[0043] FIG8A depicts a schematic diagram showing an example of a process for hybrid protein design, wherein alternate protein sequences are generated for the antibody trastuzumab, according to some exemplary embodiments;

[0044] FIG8B depicts a graph showing human epidermal growth factor receptor 2 (HER2) binding affinities exhibited by alternate protein sequences and structures generated by various hybrid protein design workflows for the antibody trastuzumab, according to some exemplary embodiments;

[0045] FIG8C depicts a graph showing antigen-binding fragment (Fab) stability of alternate protein sequences and structures generated for the antibody trastuzumab by various hybrid protein design workflows according to some exemplary embodiments; and

[0046] FIG. 9 depicts a block diagram illustrating an example of a computing system in accordance with some exemplary embodiments.

[0047] When applicable, like reference numerals refer to like structures, features, or elements. DETAILED DESCRIPTION

[0048] Protein design aims to identify novel protein sequences (e.g., sequences of amino acid residues) that exhibit certain desired properties, including, for example, expression, binding affinity for a target molecule (e.g., an antigen such as a viral antigen or a tumor antigen), specificity for a target molecule, lack of nonspecificity, stability (e.g., conformational stability, thermodynamic stability, robustness to different environmental stresses (such as protease resistance), etc.), non-immunogenicity, human nature, absence of self-association (or non-aggregation), lack of chemical propensity (e.g., aspartate isomerization, oxidation, deamidation), developability, etc. However, protein design is a challenging and particularly resource-intensive task, at least due to the countless variations in protein sequence and structure. For example, the sequence space containing possible arrangements of amino acid residues that can form a protein molecule is enormous (e.g., for For a protein sequence of amino acid residues, At the same time, even when restricted to a few discrete (rather than continuous) structural variants, the conformational space filled by the possible three-dimensional structures of a protein molecule is similarly extremely large. For example, if each amino acid residue is restricted to assuming one of three discrete geometric states (e.g., rotamers), then a protein sequence of N amino acid residues has approximately 3 N possible conformations. As used herein, the term "space" refers to the solution space for a given problem. In the context of protein sequence design, the aforementioned sequence space may represent possible solutions associated with constructing a protein sequence (of specified or unspecified length) using a set of amino acid residues (such as the amino acid residues that form an antibody). In the case of protein structure design, the aforementioned conformational space may represent possible solutions associated with determining the geometric states of the amino acid residues that form the protein sequence (including, for example, the geometric states of the backbone atoms and side chain atoms in each amino acid residue that forms the protein sequence).

[0049] Therefore, the structurally agnostic prediction of protein sequences can produce many proteins that cannot adopt a three-dimensional structure that can bind to the target molecule. Predicting only protein structure can provide many possible protein structures, which can structurally bind to the target antigen, but lack other desired properties (e.g., expression, specificity for the target molecule, lack of non-specificity, stability, non-immunogenicity, humanity, absence of self-association (or non-aggregation), lack of chemical tendency, etc.). Both the sequence space and the conformational space are so huge that many proteins predicted by computational means may have almost no therapeutic value because the predicted protein sequence lacks a structure complementary to the target molecule and / or other desired properties. Therefore, the separate design method including the application of a protein structure computational model to the input protein sequence only determines the three-dimensional structure adopted by the protein sequence, but the protein sequence itself may still be suboptimal. For example, a protein sequence generated independently (e.g., generated by a language-based protein sequence computational model) may fail to exhibit certain desired properties, such as binding affinity and stability, at least because the language-based protein sequence computational model lacks the recognition of structural features that contribute to these desired properties.

[0050] The present disclosure explains the suboptimal results produced by independently determining the sequence and structure of protein molecules. The present system and method recognize that the three-dimensional structure and function of a protein molecule depends on the sequence of amino acid residues that form the protein molecule (e.g., the primary structure of the protein molecule). For example, the binding affinity between a protein molecule and a target molecule may depend on whether the primary structure of the protein molecule can adopt a three-dimensional structure (e.g., secondary structure and tertiary structure) that is complementary to the three-dimensional structure of the target antigen. In some cases, the actual utility of a protein molecule as a therapeutic agent may further depend on the stability of its three-dimensional structure or its part (e.g., antigen binding fragment (Fab), etc.). Therefore, determining a protein sequence that is more likely to exhibit certain desired properties may include a joint search across sequence space and conformational space to identify, for example, an arrangement of amino acid residues that can present a specific three-dimensional structure. Therefore, in some exemplary embodiments, the present disclosure provides a protein design engine that can perform such a hybrid protein design workflow that integrates protein sequence design and protein structure design so that the generation of the sequence of amino acid residues that form the protein molecule is informed by the three-dimensional structure of the protein molecule. However, naive hybrid protein design methods that combine brute force searches of sequence space and conformational space are computationally intractable for protein sequences of meaningful length. Therefore, as described in more detail below, protein design engines can utilize protein sequences generated by protein sequence computational models, not as independent possible therapeutic proteins, but as sequences used to refine or guide the sequence space and conformational space that are searched by protein structure computational models when generating sequences and three-dimensional structures of protein molecules. Doing so can reduce the computational burden associated with the hybrid protein design workflow, while the resulting protein sequences can be more likely to exhibit one or more desired properties than those generated by protein sequence models operating without any structural knowledge.

[0051] In some exemplary embodiments, the protein design engine may perform a hybrid protein design workflow in which the protein structure computational model refines the searched sequence space and conformational space based at least on the protein sequence generated by the protein sequence computational model to generate the sequence and three-dimensional structure of the protein molecule. For example, in some cases, the protein design engine may apply the protein sequence computational model to generate multiple proposed protein sequences based at least on the input protein sequence. The protein sequence computational model may require a language-based protein sequence computational model, etc. In this context, the language-based protein sequence computational model may refer to a machine learning model (or deep learning model) that applies one or more natural language processing (NLP) techniques to generate a protein sequence. Examples of machine learning architectures for language-based protein sequence computational models include autoencoders, converters, long short-term memory networks, recursive neural networks, and the like.

[0052] In some cases, the input protein sequence may be selected as a basis for generating the plurality of proposed protein sequences at least because the input protein sequence exhibits one or more desired characteristics. The protein design engine may reduce the sequence space and conformation space searched by the protein structure computational model by at least the following: identifying a set of possible amino acid residues for each position in at least a portion of the output protein sequence based at least on the plurality of proposed protein sequences. As described in more detail below, the protein design engine may utilize the set of possible amino acid residues for each position in the output protein sequence to exclude at least one possible amino acid residue at one or more positions in the output protein sequence by using various means to reduce the sequence space and conformation space searched by the protein structure computational model. In doing so, the possible arrangements of amino acid residues that form the output protein sequence and the amount of their corresponding conformations may be reduced.

[0053] In some cases, the protein design engine can reduce the sequence space and conformation space searched by the protein structure computational model by applying a property analysis model. The property analysis model of the present disclosure can determine the properties of each protein sequence in the multiple proposed protein sequences generated by the protein sequence computational model. In some cases, at least one protein sequence can be excluded from the multiple proposed protein sequences based on its properties. The amount of possible amino acid residues for one or more positions in the output protein sequence can then be reduced by simply identifying the set of possible amino acid residues for each position in the output protein sequence based on the remaining multiple proposed protein sequences. Doing so can further reduce the sequence space and conformation space searched by the protein structure computational model.

[0054] In addition, in some cases, the plurality of proposed protein sequences can be aligned to identify the amino acid residue that occurs at each position across the aligned plurality of proposed protein sequences. For example, in some cases, the plurality of proposed protein sequences generated by a protein sequence computational model can be aligned by applying one or more alignment techniques (such as dynamic programming, progressive alignment, hierarchical alignment, iterative alignment, motif discovery, deep learning models, hidden Markov models, etc.).

[0055] In some exemplary embodiments, the protein design engine may determine, for each position in at least a portion of the output protein sequence, a set of possible amino acid residues that includes a first amino acid residue but not a second amino acid residue. For example, in some cases, the set of possible amino acid residues for the position in the output protein sequence may be identified as including the first amino acid residue based at least on a first frequency at which the first amino acid residue appears at a particular position across the multiple proposed protein sequences generated by the protein sequence computational model. Alternatively and / or in addition, the set of possible amino acid residues for the position may be identified as excluding the second amino acid residue based at least on a second frequency at which the second amino acid residue appears at the position across the multiple proposed protein sequences generated by the protein sequence computational model. Thus, in some cases, the set of possible amino acid residues for the position may be determined to include some but not all of the possible amino acid residues that can form a protein molecule. For example, in the case where the output protein sequence corresponds to an antibody, the set of amino acid residues for that position may include some, but not all, of the amino acid residues that can form an antibody (which may include, for example, alanine, arginine, asparagine, aspartic acid, cysteine, glutamic acid, glutamine, glycine, histidine, isoleucine, leucine, lysine, methionine, phenylalanine, proline, serine, threonine, tryptophan, tyrosine, valine, etc.).

[0056] In some exemplary embodiments, the protein design engine may generate a first protein structure having an output protein sequence by at least applying a protein structure computational model to select amino acid residues from the set of possible amino acid residues for inclusion in the output protein sequence for each position in at least the portion of the output protein sequence. For example, in some cases, the protein structure computational model may generate a first protein structure by at least determining a plurality of amino acid residues for inclusion in the output protein sequence and corresponding conformations (or spatial arrangements) of the plurality of amino acid residues (e.g., backbone and side chain atoms of each amino acid residue) based at least on one or more criteria. As noted, each position in the output protein sequence may be associated with a set of possible amino acid residues, which in some cases contain a subset of each amino acid residue that can form an antibody. Thus, the selection of amino acid residues that form the output protein sequence may be limited to those amino acid residues that are included in each possible amino acid residue set. In some cases, the selection of amino acid residues that form the output protein sequence may be limited, wherein one or more segments of the output protein sequence are designated for retention. In this context, retention of a segment in the input protein sequence may require preventing any insertion, deletion, or substitution from affecting the segment when generating the output protein sequence. Thus, it will be appreciated that the identity of the amino acid residues in the retained segments of the output protein sequence remains unchanged compared to the identity of the amino acid residues occupying the same position in the input protein sequence. That is, the same amino acid residues that form the retained segments in the input protein sequence will remain present in the retained segments in the output protein sequence. Thus, for a position in the retained segments of the output protein sequence, the set of possible amino acid residues for that position may contain a single amino acid residue occupying the same position in the input protein sequence.

[0057] In some cases, the one or more criteria may include minimizing an energy function of a first protein structure having the plurality of amino acid residues selected for inclusion in the output protein sequence. In addition, in some cases, the protein structure computational model may generate a plurality of protein molecules (each of which has a different amino acid residue sequence and / or conformation) and then select a protein molecule that meets the one or more criteria. The reduction in sequence space and conformational space searched by the protein structure computational model may be reflected in a sharp reduction in the amount of possible protein molecules (e.g., having different combinations of amino acid residues and conformations) evaluated by the protein structure computational model. For further demonstration, the protein structure computational model may determine, for example, a first energy of a first protein molecule having a first plurality of amino acid residues and a first conformation. In addition, the protein structure computational model may generate a second protein molecule having a second plurality of amino acid residues and a second conformation by modifying at least one of the first plurality of amino acid residues and the first conformation. The protein structure computational model may determine a second energy of the second protein molecule. In the case where the first energy of the first protein molecule is lower than the second energy of the second protein molecule, the protein structure computational engine may generate the first protein structure as having the first plurality of amino acid residues and the first conformation (rather than the second plurality of amino acid residues and the second conformation). In some cases, the protein structure computational model can determine the first energy and the second energy by applying an energy function. For example, the protein structure computational model can apply an energy function based on one or more of the following: ab initio quantum mechanics, density functional theory (DFT), semi-empirical methods, molecular mechanics force fields, statistical potentials, neural potentials, machine learning models (e.g., trained on structural data), etc.

[0058] In some exemplary embodiments, the protein structure computational model may further generate the first protein structure by at least determining a first backbone conformation of a first backbone of the first protein structure. The first backbone of the first protein structure may be a continuous atomic chain formed by connecting the backbone atoms of each amino acid residue in the output protein sequence. In some cases, the backbone atoms of each amino acid residue may be a nitrogen (N) atom, atom and a carboxyl carbon (C) atom. Thus, in some cases, the first backbone conformation of the first backbone may include a spatial arrangement of backbone atoms selected for inclusion in each amino acid residue in the output protein sequence. Furthermore, in some cases, the spatial arrangement of backbone atoms may be defined by one or more of the following: a translation of the first backbone, a rotation of the first backbone, and / or a torsion angle of one or more rotatable bonds formed by backbone atoms in the first backbone (e.g., The torsion angle ψ of the rotatable bond between the atom and the carbonyl group, The torsion angle ϕ of the rotatable bond between the carbon (C) atom and the nitrogen (N) atom, the torsion angle ω of the rotatable bond between the carbon (C) atom and the nitrogen (N) atom, etc.).

[0059] In some exemplary embodiments, the protein structure computational model may determine that the first backbone conformation of the first skeleton of the first protein structure has the same conformation as at least a portion of the second skeleton of the second protein structure. In some cases, the second protein structure may be associated with a protein sequence having one or more desired properties (such as expression, binding affinity for a target molecule, specificity for a target molecule, lack of non-specificity, stability, non-immunogenicity, human nature, absence of self-association (or non-aggregation), etc.). For example, in some cases, the protein sequence may be an input protein sequence based on which the protein sequence computational model generates the plurality of proposed protein sequences or the third protein sequence. In addition, in some cases, the second skeleton of the second protein structure may exhibit the second backbone conformation of the second protein structure in an unbound state or the third backbone conformation of the second protein structure bound to the target molecule.

[0060] In some exemplary embodiments, rather than the first backbone conformation of the first skeleton being determined to have the same conformation as at least the portion of the second skeleton of the second protein structure, the protein structure computational model can determine the first backbone conformation of the first skeleton by at least determining the geometric state of the first skeleton. In some cases, the geometric state of the first skeleton can be defined by the translation and / or rotation of the skeleton atoms contained in each amino acid residue in the output protein sequence, so that the protein structure computational model determines the geometric state of the first skeleton by at least determining the translation and / or rotation of the skeleton atoms. Alternatively and / or in addition, the geometric state of the first skeleton can be defined by the torsion angle of one or more rotatable bonds formed by the skeleton atoms in each amino acid residue in the output protein sequence. Therefore, in some cases, the protein structure computational model can determine the geometric state of the first skeleton by at least determining the torsion angle of the one or more rotatable bonds formed by the skeleton atoms.

[0061] In some exemplary embodiments, after determining the first backbone conformation of the first backbone of the first protein structure, the protein design computational model may further generate the first protein structure by at least determining that at each position in at least a portion of the first backbone having the first backbone conformation, an amino acid residue selected for inclusion at the corresponding position in the output protein sequence is included. In addition, the protein structure computational model may generate the first protein structure by at least determining, for each position in at least a portion of the first backbone having the first backbone conformation, a side chain conformation of an amino acid residue selected for inclusion at the corresponding position in the output protein sequence. For example, in some cases, the side chain conformation of the amino acid residue may be determined based at least on the first backbone conformation of the first backbone of the first protein structure. In addition, the protein structure computational model may determine the side chain conformation of the amino acid residue selected for inclusion in the output protein sequence by at least determining a geometric state of the side chain. For example, in some cases, the geometric state of the side chain of an amino acid residue can be determined by at least the following: selecting a rotational isomer from a plurality of possible rotational isomers, each of the plurality of possible rotational isomers comprising a different combination of torsion angles of rotatable bonds formed by side chain atoms in the amino acid residue, alternatively, the geometric state of the side chain can be determined by determining one or more of the translation, rotation and / or torsion angles of the side chain atoms in the amino acid residue.

[0062] In some exemplary embodiments, the protein design engine may validate the first protein structure generated by the protein structure computational model by at least applying a different protein structure computational model to determine a second protein structure having at least the output protein sequence based at least on the output protein sequence. In addition, the protein design engine may validate the first protein structure generated by the protein structure computational model by determining a similarity metric (e.g., root mean square deviation (RMSD), etc.) that quantifies the difference between the first protein structure and the second protein structure. In the case where the similarity metric satisfies one or more thresholds, the protein design engine may validate the first protein structure and identify the corresponding output protein sequence as a candidate for synthesis, in vitro measurement, in vivo characterization, etc.

[0063] In some exemplary embodiments, the protein design engine may generate a first protein structure such that one or more portions of the first protein structure may be used as a donor structure for grafting onto a corresponding portion of a second protein structure. For example, in some cases, the second protein structure may be selected at least because the second protein structure exhibits one or more desired properties, such as binding affinity for a target molecule, specificity for a target molecule, lack of non-specificity, stability, non-immunogenicity, human nature, absence of self-association (or non-aggregation), etc. Thus, in some cases, the protein design engine may identify a first portion of the second protein structure and then generate a third protein structure by replacing the first portion of the second protein structure with at least a second portion of the first protein structure. In some cases, the protein design engine may verify the third protein structure based at least on a similarity metric (e.g., root mean square deviation (RMSD), etc.) that quantifies the difference between the third protein structure and at least a fourth protein structure generated by a protein structure computational model and / or a different protein structure computational model based on a third protein sequence of the third protein structure. In the case where the first protein structure and the second protein structure are antibodies, the first portion of the second protein structure may include a first antigen binding site (e.g., a first paratope, a first complementarity determining region (CDR), etc.), and the second portion of the first protein structure may include a second antigen binding site (e.g., a second paratope, a second complementarity determining region (CDR), etc.). Thus, grafting the second portion of the first protein structure onto the second protein structure may generate an antibody whose sequence and structure incorporate the sequences and structures of multiple antibodies.

[0064] FIG1 depicts a system diagram showing an example of a protein design system 100 according to some exemplary embodiments. Referring to FIG1 , the protein design system 100 may include a protein design engine 110, a molecular analysis engine 120, and a client device 130. As shown in FIG1 , the protein design engine 110, the analysis engine 120, and the client device 130 may be communicatively coupled via a network 140. The client device 130 may be a processor-based device, including, for example, a workstation, a desktop computer, a laptop computer, a smartphone, a tablet computer, a wearable device, etc. The network 140 may be a wired network and / or a wireless network, including, for example, a local area network (LAN), a virtual local area network (VLAN), a wide area network (WAN), a public land mobile network (PLMN), the Internet, etc.

[0065] In some exemplary embodiments, the protein design engine 110 may apply the protein sequence computational model 113 to generate a plurality of proposed protein sequences 155 based on the input protein sequence 150a, including, for example, a first proposed protein sequence 155a, a second proposed protein sequence 155b, a third proposed protein sequence 155c, etc. In some cases, the protein design engine 110 may receive the input protein sequence 150a including an amino acid residue sequence from the client device 130. Furthermore, in some cases, the input protein sequence 150a may be selected as a basis for generating the plurality of proposed protein sequences 155 at least because the input protein sequence 150a exhibits one or more desired properties, including, for example, expression, binding affinity for a target molecule (e.g., an antigen, such as a viral antigen or a tumor antigen), specificity for a target molecule, lack of non-specificity, stability, non-immunogenicity, human nature, absence of self-association (or non-aggregation), etc.

[0066] In some exemplary embodiments, the protein sequence computational model 113 may include one or more machine learning models trained to generate the plurality of proposed protein sequences 155. For example, in some cases, the one or more machine learning models may generate each of the plurality of proposed protein sequences 155 by at least sampling a data distribution learned by the one or more machine learning models during training based on the input protein sequence 150a. That is, each sampling of the data distribution may correspond to a single sampling iteration that generates a single protein sequence in the plurality of proposed protein sequences 155. For example, a first sampling of the data distribution may generate a first proposed protein sequence 155a, while a second sampling of the data distribution may generate a second proposed protein sequence 155b. In some cases, the one or more machine learning models may be trained based on a plurality of known protein sequences, including protein sequences known to exhibit certain functions and protein sequences without any known functions. Thus, the one or more machine learning models may be trained to learn a data distribution corresponding to a dimensionality-reduced representation of a sequence of amino acid residues that forms a known protein sequence.

[0067] In some cases, the one or more machine learning models may include an autoencoder (e.g., a denoising autoencoder (DAE), etc.), in which case the one or more machine learning models may learn the data distribution by at least learning to generate an encoding of an input protein sequence that is then decoded to form an output protein sequence that is minimally different from the input protein sequence. At inference time, the data distribution associated with the trained autoencoder may be sampled by encoding, for example, the input protein sequence 150a, and then decoding an intermediate sequence that has at least one of a disruption (e.g., an insertion, deletion, and / or modification of amino acid residues) and a length variation relative to the input protein sequence 150a. In addition, the sampling of the data distribution may be guided by the characteristics of the intermediate sequence. For example, in some cases, the intermediate sequence sampled from the data distribution may be subjected to a characteristic analysis (e.g., by computing a function prediction model), and if the intermediate sequence is determined to exhibit one or more desired characteristics that may also be present in the input protein sequence 150a, the intermediate sequence may be decoded. In the case where the intermediate sequence fails to exhibit the one or more desired properties, another intermediate sequence may be generated by sampling the data distribution before the intermediate sequence is also subjected to property analysis. Thus, the intermediate sequence decoded to generate each of the plurality of proposed protein sequences 155 may be different from the input protein sequence 150a, but will still exhibit the same (or similar) desired properties as the input protein sequence 150a. For example, the first proposed protein sequence 155a, the second proposed protein sequence 155b, and the third proposed protein sequence 155c may be different from the input protein sequence 150a, but will retain at least some of the desired properties of the input protein sequence 150a.

[0068] In some exemplary embodiments, the protein design engine 110 may apply a variety of sampling techniques (including, for example, Markov Monte Carlo (MCMC), importance sampling (IS), rejection sampling, Metropolis-Hastings, Gibbs sampling, slice sampling, exact sampling, etc.) to sample from the data distribution. Moreover, as noted, each sampling of the data distribution may correspond to a single sampling iteration that generates one of the plurality of proposed protein sequences 155. In some cases, the protein design engine 110 may continue to apply the protein sequence computational model 113 to sample the data distribution until one or more conditions are met. For example, in some cases, the protein design engine 110 may continue to apply the protein sequence computational model 113 to sample the data distribution until the plurality of proposed protein sequences 155 generated by the protein sequence computational model 113 includes a threshold amount of proposed protein sequences. Alternatively and / or in addition, each proposed protein sequence in the plurality of proposed protein sequences 155 generated by the protein sequence computational model 113 may be subjected to molecular analysis (e.g., by applying the property analysis model 125 by the molecular analysis engine 120) in order to determine one or more properties of each proposed protein sequence. In those cases, the protein design engine 110 may continue to apply the protein sequence computational model 113 to generate additional proposed protein sequences until the plurality of proposed protein sequences 155 generated by the protein sequence computational model 113 includes a threshold amount of proposed protein sequences whose properties satisfy one or more thresholds. Thus, in some cases, the plurality of proposed protein sequences 155 may exclude one or more proposed protein sequences whose properties (e.g., as determined by the property analysis model 125) fail to satisfy the one or more thresholds.

[0069] In some exemplary embodiments, the protein design engine 110 may perform variable length sampling, wherein the plurality of proposed protein sequences 155 undergoing further analysis have different lengths or amounts of constituent amino acid residues. For example, in some cases, the first proposed sequence 155a may have a different length (or be formed by a different amount of amino acid residues) compared to the second proposed sequence 155b and / or the third proposed sequence 155c. Alternatively, in some cases, the protein design engine 110 may perform fixed length sampling, wherein the plurality of proposed protein sequences 155 (including the first proposed sequence 155a, the second proposed sequence 155b, and the third proposed sequence 155c, etc.) have the same length (or are formed by the same amount of amino acid residues). In some cases, the protein design engine 110 may implement fixed length sampling by at least the following: training the protein sequence computational model 113 based on a training data set containing protein sequences of the same length. In doing so, the data distribution learned by the protein sequence computational model 113 may be populated by encodings of protein sequences that decode to protein sequences of the same length. Therefore, the plurality of proposed protein sequences 155 generated by applying the protein sequence computational model 113 may also have the same length (or the same amount of constituent amino acid residues). In cases where the protein sequence computational model 113 is trained based on protein sequences of different lengths, the protein sequence computational model 113 may generate protein sequences with different lengths (or different amounts of constituent amino acid residues). In those cases, the protein design engine 110 may implement fixed-length sampling by at least excluding from the plurality of proposed protein sequences 155 any protein sequence generated by the protein sequence computational model 113 that does not have a specific length (or a specific amount of constituent amino acid residues).

[0070] Referring again to FIG. 1 , in some exemplary embodiments, the protein design engine 110 may include a sequence analyzer 115 that determines a possible set of amino acid residues 160 for each position in the output protein sequence 150b based at least on the plurality of proposed protein sequences 155 generated by the protein sequence computational model 113. As noted, in the case where the protein design engine 110 performs fixed-length sampling, the plurality of proposed protein sequences 155 may include protein sequences having the same length (or the same amount of constituent amino acid residues), while in the case of variable-length sampling, the plurality of proposed protein sequences 155 may differ in length (or amount of constituent amino acid residues). In some cases, the sequence analyzer 115 may align the plurality of proposed protein sequences 155 to identify the amino acid residue that occurs at each position across the aligned plurality of proposed protein sequences 157. For example, in some cases, the sequence analyzer 115 may generate the aligned plurality of proposed protein sequences 157 by at least applying one or more of dynamic programming, progressive alignment, hierarchical alignment, iterative alignment, motif discovery, deep learning models, hidden Markov models, etc. to the plurality of proposed protein sequences 155. As described in more detail below, the possible amino acid residue sets 160 for at least some of the positions in the output protein sequence 150b may include a subset of amino acid residues containing some but not all of the possible constituent amino acid residues of a protein molecule. In the case where the output protein sequence 150b corresponds to an antibody, the possible amino acid residue sets 160 may include some but not all of the amino acid residues that may form an antibody (such as, for example, alanine, arginine, asparagine, aspartic acid, cysteine, glutamic acid, glutamine, glycine, histidine, isoleucine, leucine, lysine, methionine, phenylalanine, proline, serine, threonine, tryptophan, tyrosine, valine, etc.).

[0071] In some exemplary embodiments, the protein design engine 110 may apply the protein structure computational model 117 to generate a first protein structure 170 by at least selecting, for each position in at least a portion of the output protein sequence 150b, an amino acid residue from a corresponding set of possible amino acid residues 160 for inclusion in the output protein sequence 150b. In some cases, the protein structure computational model 117 may include a physics-based protein structure computational model, a machine learning-based protein structure computational model, and the like. As will be described in more detail below, the protein structure computational model 117 may generate the first protein structure 170 by at least determining a first backbone conformation of a first backbone of the first protein structure 170. Furthermore, the protein structure computational model 117 may generate the first protein structure 170 by at least determining, for each position in at least a portion of the first backbone having the first backbone conformation, a side chain conformation of an amino acid residue selected for inclusion in the output protein sequence 150b at the corresponding position.

[0072] 2 depicts a flow chart showing an example of a process 900 for hybrid protein design according to some exemplary embodiments. Referring to FIGS. 1 and 2 , process 900 may be performed by protein design engine 110 to generate output protein sequence 150b and corresponding first protein structure 170 based at least on input protein sequence 150a.

[0073] At 202, the protein design engine 110 may apply the protein sequence computational model 113 to generate a plurality of proposed protein sequences based at least on the input protein sequence. In some exemplary embodiments, the protein sequence computational model 113 may generate the plurality of proposed protein sequences 155 based on the input protein sequence 150a, including, for example, a first proposed protein sequence 155a, a second proposed protein sequence 155b, a third proposed protein sequence 155c, and the like. In some cases, the protein sequence computational model 113 may generate the plurality of proposed protein sequences 155 by sampling at least a data distribution populated by encodings of various protein sequences. The data distribution may be learned by the protein sequence computational model 113 by at least the following: training the protein sequence computational model 113 to encode known protein sequences (including protein sequences known to exhibit certain functions and protein sequences without any known functions) so that the known protein sequences can be restored (e.g., by decoding) from the corresponding encodings with minimal information loss (e.g., the difference between the decoded protein sequence and the original protein sequence). For each sampling of the data distribution, the protein sequence computational model 113 can generate a single proposed protein sequence that is different from the input protein sequence 150a, for example, by having at least one of a disruption (e.g., an insertion, deletion, and / or modification of an amino acid residue) and a length variation relative to the input protein sequence 150a. As noted, the sampling of the data distribution can be guided by the properties of the intermediate sequences, so that the plurality of proposed sequences 155 generated therefrom can exhibit certain desired properties, including, in some cases, properties that are the same (or similar) as the input protein sequence 150a.

[0074] At 204, the protein design engine 110 may identify a set of possible amino acid residues for each position in at least a portion of the output protein sequence based at least on the plurality of proposed protein sequences. In some exemplary embodiments, the sequence analyzer 115 may determine a possible set of amino acid residues 160 for each position in the output protein sequence 150b based at least on the plurality of proposed protein sequences 155. As shown in FIG. 1 , in some cases, the sequence analyzer 115 may align the plurality of proposed protein sequences 155, for example, by applying one or more alignment techniques, including, for example, dynamic programming, progressive alignment, hierarchical alignment, iterative alignment, motif discovery, deep learning models, hidden Markov models, etc. In doing so, the sequence analyzer 115 may generate an aligned proposed sequence 157, and then determine a possible set of amino acid residues 160 for each position in the output protein sequence 150b based at least on the aligned proposed sequence 157.

[0075] For further illustration, FIG6A depicts a schematic diagram showing an example of an aligned proposed sequence 157 generated by the sequence analyzer 115 aligning at least a first proposed sequence 155a, a second proposed sequence 155b, and a third proposed sequence 155c. In some exemplary embodiments, once the plurality of proposed protein sequences 155 are aligned, the sequence analyzer 115 may identify the amino acid residue that occurs at each position across the aligned proposed sequences 157. Furthermore, in some cases, the sequence analyzer 115 may determine the frequency with which each amino acid residue occurs at each position across the aligned proposed sequences 157. 6A, once the first proposed sequence 155a, the second proposed sequence 155b, and the third proposed sequence 155c are aligned, the sequence analyzer 115 can determine that the amino acid residue arginine (R) occurs in the first position 600a of the first proposed sequence 155a, while the amino acid residue lysine (K) occurs at the first position 600a of the second proposed sequence 155b, and the amino acid residue glutamine (Q) occurs in the first position 600a of the third proposed sequence 155c. The sequence analyzer 115 can also determine that the amino acid residue alanine (A) occurs in the second position 600b of the first proposed sequence 155a, the second proposed sequence 155b, and the third proposed sequence 155c. Furthermore, the sequence analyzer 115 may determine that the amino acid residue serine (S) occurs in the third position 600c of the first proposed sequence 155a and the third proposed sequence 155c, and the amino acid residue threonine (T) occurs in the third position 600c of the second proposed sequence 155b.

[0076] In some exemplary embodiments, the sequence analyzer 115 may determine amino acid residues for which corresponding likelihoods of appearing in the output protein sequence 150b meet one or more thresholds based at least on the aligned proposed sequences 157. For example, in some cases, the sequence analyzer 115 may identify the amino acid residue that is most likely to occupy each position in at least a portion of the output protein sequence 150b based at least on the amino acid residues occupying each position across the aligned proposed sequences 157. Thus, in some cases, the sequence analyzer 115 may generate a possible set of amino acid residues 160 for each position to exclude at least some of the amino acid residues that can form a protein molecule. Doing so may reduce the sequence space of possible arrangements of amino acid residues from the protein structure computational model 117 that determines the output protein sequence 150b and the corresponding protein structure 170 during subsequent structural design.

[0077] In some exemplary embodiments, the sequence analyzer 115 may determine a possible set of amino acid residues 160 for each position in at least a portion of the output protein sequence 150b based at least on the amino acid residues that appear in each position across the aligned proposed sequences 157. For example, the possible set of amino acid residues 160 for a particular position may be identified as including amino acid residues that are more likely to occupy that position in the output protein sequence 150b and excluding amino acid residues that are less likely to occupy that position in the output protein sequence 150b. In the case where the output protein sequence 150b is an antibody, the possible set of amino acid residues 160 for that position may include a subset of amino acid residues that may form an antibody. For example, where the output protein sequence 150b is an antibody, the sequence analyzer 115 may determine some, but not all, of the following amino acid residues based at least on the amino acid residue occurring in each position across the aligned proposed sequences 157: alanine, arginine, asparagine, aspartic acid, cysteine, glutamic acid, glutamine, glycine, histidine, isoleucine, leucine, lysine, methionine, phenylalanine, proline, serine, threonine, tryptophan, tyrosine, and valine.

[0078] In some cases, one or more amino acid residues may be identified for inclusion in the possible amino acid residue set 160 for a particular position in the output protein sequence 150b based at least on the frequency with which each amino acid residue occurs at that position. For example, if the frequency with which a first amino acid residue occurs in a position across the aligned proposed sequences 157 indicates that the first amino acid residue is likely to occupy that position in the output protein sequence 150b, then the first amino acid residue may be identified for inclusion in the possible amino acid residue set 160 for that position. Conversely, if the frequency with which a second amino acid residue occurs in the position across the aligned proposed sequences 157 indicates that the second amino acid residue is unlikely to occupy that position in the output protein sequence 150b, then the second amino acid residue may be excluded from the possible amino acid residue set for that position. Thus, in some cases, if the frequency with which the one or more amino acid residues occur at a position across the aligned proposed sequences 157 satisfies one or more thresholds, then the one or more amino acid residues may be identified for inclusion in the possible amino acid residue set 160 for that position in the output protein sequence 150b. In some cases, the one or more thresholds may be determined based on the maximum, minimum, median, mean, and mode of the frequencies at which each amino acid residue in the plurality of amino acid residues appears at the position across the aligned proposed sequence 157. For example, in some cases, if a first frequency at which a first amino acid residue appears in a particular position in the output protein sequence 150b across the aligned proposed sequence 157 exceeds the median frequency at which each amino acid residue in the plurality of amino acid residues appears at the position across the aligned proposed sequence 157, the first amino acid residue may be identified for inclusion in the set of possible amino acid residues 160 at the position. In contrast, if a second frequency at which a second amino acid residue appears in the position across the aligned proposed sequence 157 fails to exceed the aforementioned median frequency, the second amino acid residue may be excluded from the set of possible amino acid residues 160 at the position.

[0079] For further illustration, FIG6B depicts a schematic diagram showing an example of possible amino acid residue groups 160 generated for each position in at least a portion of an output protein sequence 150b. In the example shown in FIG6B, a first position 600a of the output protein sequence 150b may be associated with a first possible amino acid residue group 160a including amino acid residues arginine (R), lysine (K), glycine (Q), and glutamine (Q). The sequence analyzer 115 may select amino acid residues arginine (R), lysine (K), glycine (Q), and glutamine (Q) for inclusion in the first possible amino acid residue group 160a at least because the frequency with which amino acid residues arginine (R), lysine (K), glycine (Q), and glutamine (Q) appear in the first position 160a across the aligned proposed sequence 157 satisfies one or more thresholds. At the same time, the second position 600b of the output protein sequence 150b may be associated with a second possible amino acid residue group 160b including the amino acid residue alanine (A) (which may be selected for inclusion based at least on the frequency of the amino acid residue alanine (A) appearing in the second position 600b across the aligned proposed sequence 157 satisfying one or more thresholds). In addition, the third position 600c of the output protein sequence 150b may be associated with a third possible amino acid residue group 160c including the amino acid residues serine (S) and threonine (T). The amino acid residues serine (S) and threonine (T) may be included in the third possible amino acid residue group 160c of the third position 600c at least because the frequency of these amino acid residues appearing in the third position 600c across the aligned proposed sequence 157 satisfies one or more thresholds. It should be understood that amino acid residues not present in each possible amino acid residue group 160 may be excluded at least because the frequency of these amino acid residues appearing at the corresponding position across the aligned proposed sequence 157 fails to satisfy one or more thresholds.

[0080] At 206, the protein design engine 110 may generate a protein structure with the output protein sequence by at least applying the protein structure computational model 117 to select, for each position in at least a portion of the output protein sequence, an amino acid residue from a corresponding set of possible amino acid residues for inclusion in the output protein sequence. In some exemplary embodiments, the protein design engine 110 may apply the protein structure computational model 117, which may generate a first protein structure 170 with the output protein sequence 150b by at least selecting, for each position in at least a portion of the output protein sequence 150b, an amino acid residue from a corresponding set of possible amino acid residues 160 for inclusion in the output protein sequence 150b. The protein structure computational model 117 operates on the set of possible amino acid residues 160 for each position (which contains some but not all of the possible amino acid residues that form the protein molecule), which reduces the computational complexity of the structure design performed by the protein structure computational model 117. For example, the protein structure computational model 117 can generate a first protein structure 170 having an output protein sequence 150b by at least evaluating various possible protein molecules, each of which has a different arrangement of amino acid residues and conformation. Therefore, in order to generate the first protein structure 170 having the output protein sequence 150b, the protein structure computational model 117 can perform a joint search across the sequence space of possible arrangements of amino acid residues forming protein molecules and the conformation space of the corresponding three-dimensional structure. The computational burden of the search can be reduced by: the protein structure computational model 117 utilizes the possible amino acid residue groups 160 at each position to significantly reduce the amount of possible protein molecules evaluated to generate the output protein sequence 150b.

[0081] In some cases, the protein structure computational model 117 may perform a joint search across sequence space and conformation space by at least evaluating the three-dimensional structures of different arrangements of amino acid residues formed by the possible amino acid residues in each set of possible amino acid residues 160. For example, in some cases, the protein structure computational model 117 may be applied to determine the energy of each three-dimensional protein structure in order to identify the arrangement of amino acid residues that folds into the three-dimensional structure with the minimum energy. As described in more detail below, for each arrangement of amino acid residues that forms the output protein sequence 150b, the protein structure computational model 117 may determine the corresponding first protein structure 170 by at least determining the backbone conformation of the first protein structure 170 having the particular arrangement of amino acid residues and / or the conformation of the side chains of the constituent amino acid residues. For example, in some cases, the protein structure computational model 117 may determine the geometric states of the backbone and / or the side chains of the constituent amino acid residues of the first protein structure 170 with different degrees of freedom. In the case of backbone conformation, the protein structure computational model 117 may determine the geometric state of the backbone of the first protein structure 170 based at least in part on the backbone conformation of another protein structure associated with one or more desired properties (e.g., expression, binding affinity for a target molecule (e.g., an antigen, such as a viral antigen or a tumor antigen), specificity for a target molecule, lack of non-specificity, stability, non-immunogenicity, human nature, absence of self-association (or non-aggregation), etc.). At the same time, the side chain conformation of the amino acid residues may be determined by selecting one of a plurality of possible rotational isomers or by determining the torsion angle of one or more rotatable angles formed by the side chain atoms of the amino acid residues. FIG6B shows an example of a structural design, in which the first protein structure 170 generated by the protein structure computational model 117 is in a bound state in which the first protein structure 170 is bound to the target molecule 650. However, it should be understood that the protein structure computational model 117 may generate the first protein structure 170 to be in a bound state or an unbound state.

[0082] FIG3A depicts a flow chart showing an example of a process 300 for hybrid protein design according to some exemplary embodiments. Referring to FIG1-2 and FIG3A, process 300 may be performed by protein design engine 110 to generate output protein sequence 150b and corresponding first protein structure 170 based on at least input protein sequence 150a. In some cases, process 300 may implement operation 206 of process 200, wherein protein structure computational model 117 selects constituent amino acid residues of the protein sequence and determines the corresponding protein structure by determining at least the conformation of the backbone and side chains of the protein structure.

[0083] At 302, the protein structure computational model 117 may select a plurality of amino acid residues that meet one or more criteria for inclusion in the protein sequence. In some exemplary embodiments, the protein design engine 110 may apply the protein structure computational model 117 to select, for each position in at least a portion of the output protein sequence 150b, an amino acid residue from a corresponding possible amino acid residue set 160 for inclusion in the output protein sequence 150b. Referring again to the example shown in FIG. 6B, the protein structure computational model 117 may be applied to select, for example, an amino acid residue arginine (R) from a first possible amino acid residue set 160a, an amino acid residue alanine (A) from a second possible amino acid residue set 160b, and an amino acid residue serine (S) from a third possible amino acid residue set 160c for inclusion at a corresponding position in the output protein sequence 150b. As described in more detail below, in some cases, to identify amino acid residues for inclusion in the output protein sequence 150b, the protein structure computational model 117 may select an arrangement of amino acid residues that satisfies one or more criteria from among different possible arrangements of amino acid residues formed by the amino acid residues included in the possible amino acid residue set 160 for each position in at least a portion of the output protein sequence 150b. In some cases, the one or more criteria may include an energy of the first protein structure 170 having the output protein sequence 150b, which may be determined by applying an energy function. The possible arrangements of amino acid residues are formed by the possible amino acid residue set 160 for each position in at least the portion of the output protein sequence 150b, rather than every possible amino acid residue that may form an antibody, for example, which may significantly reduce the computational burden associated with identifying an arrangement of amino acid residues for inclusion in the output protein sequence 150b such that the output protein sequence 150 and the corresponding first protein structure 170 exhibit certain desired properties.

[0084] At 304, the protein structure computational model 117 may determine a backbone conformation of a backbone of a protein structure having a protein sequence. In some exemplary embodiments, the protein structure computational model 117 may further generate the first protein structure 170 by at least determining a conformation of the backbone of the first protein structure 170. In some cases, the backbone of the first protein structure 170 may include a continuous atomic chain formed by connecting a plurality of backbone atoms from each amino acid residue in the output protein sequence 150b. Furthermore, in some cases, the plurality of backbone atoms in each amino acid residue in the output protein sequence may be a nitrogen (N) atom, an α-carbon (C-C) atom, or a carbon atom.α ) atoms and carboxyl carbon (C) atoms. It should be understood that the same multiple backbone atoms may be present in different amino acid residues. In the example of the output protein sequence 150b shown in Figure 6B, for example, the backbone of the corresponding protein structure 170 may include backbone atoms of each amino acid residue in the sequence containing arginine (R), alanine (A), serine (S), glutamine (Q), aspartic acid (D), valine (V), asparagine (N), threonine (T), alanine (A), valine (V) and alanine (A).

[0085] As noted, the protein structure computational model 117 may be applied to determine a backbone conformation of a first protein structure 170 having different degrees of freedom. For example, in some exemplary embodiments, the protein structure computational model 117 may determine the first protein structure 170 based on a backbone of another protein structure that may be associated with a protein sequence having one or more desired properties, such as expression, binding affinity for a target molecule (e.g., an antigen, such as a viral antigen or a tumor antigen), specificity for a target molecule, lack of non-specificity, stability, non-immunogenicity, human nature, absence of self-association (or non-aggregation), etc. In some cases, the other protein structure may have an input protein sequence 150a based on which the protein sequence computational model 113 generates the plurality of proposed sequences 155. Alternatively, the other protein structure may be associated with another protein sequence that is different from the input protein sequence 150a and the output protein sequence 150b. For example, in some cases, the input protein sequence 150a used to generate the plurality of proposed sequences 155 for determining the output protein sequence 150b may be associated with a first desired property, while other protein sequences associated with other protein structures that provide the backbone conformation of the first protein structure 170 may be associated with the same first desired property and / or second desired property. Thus, in some cases, the protein structure computational model 117 may determine the first backbone conformation of the backbone of the first protein structure 170 to have the same conformation as at least a portion of the backbone of the other protein structures, such that the first protein structure 170 may exhibit the same (or similar) desired properties as the other protein structures. As described in more detail below, the protein design engine 110 may perform validation to determine whether the output protein sequence 150b will fold into the first protein structure 170 having the backbone conformation of the other protein structures.

[0086] In some exemplary embodiments, the protein design engine 110 may apply the protein structure computational model 117 to determine the backbone conformation of the first protein structure 170 independently of another protein structure. Conversely, the protein structure computational model 117 may be applied to determine the geometric state of the backbone of the first protein structure 170 having different degrees of freedom. For example, in some cases, the protein structure computational model 117 may determine the geometric state of the backbone of the first protein structure 170 by at least determining the translation and / or rotation of at least a portion of the backbone atoms that form the backbone of the first protein structure 170. In some cases, in addition to or in lieu of the translation and / or rotation of at least the portion of the backbone atoms in the backbone of the first protein structure 170, the protein structure computational model 117 may determine the geometric state of the backbone of the first protein structure 170 by at least determining the torsion angle of one or more rotatable bonds formed by at least a portion of the backbone atoms. For example, in some cases, the geometric state of the backbone of the first protein structure 170 may be determined by determining one or more of the following: The torsion angle ψ of the rotatable bond between the atom and the carbonyl group, The torsion angle ϕ of the rotatable bond between the carbon (C) atom and the nitrogen (N) atom, the torsion angle ω of the rotatable bond between the carbon (C) atom and the nitrogen (N) atom, etc.

[0087] At 306, the protein structure computational model 117 may generate the protein structure to include, at each position in at least a portion of the backbone, an amino acid residue selected for inclusion at a corresponding position in the protein sequence. In some exemplary embodiments, the protein design engine 110 may apply the protein structure computational model 117 to further generate the first protein structure 170 by at least including, at each position in at least a portion of the backbone of the first protein structure 170, side chain atoms of an amino acid residue selected for inclusion in a corresponding position in the output protein sequence 150b. In the example shown in FIG. 6B , for example, the protein structure computational model 117 may generate the first protein structure 170 by at least including, in the backbone of the first protein structure 170, side chain atoms of each of the amino acid residues arginine (R), alanine (A), serine (S), glutamine (Q), aspartic acid (D), valine (V), asparagine (N), threonine (T), alanine (A), valine (V), and alanine (A). As described in more detail below, the first protein structure 170 may be further determined by the protein design engine 110 applying the protein structure computational model 117 to determine the side chain conformation of each amino acid residue selected for inclusion in the output protein sequence 150b.

[0088] At 308, the protein structure computational model 117 may determine a side chain conformation of each amino acid residue selected for inclusion in the output protein sequence. In some exemplary embodiments, the protein design engine 110 may apply the protein structure computational model 117 to further generate a first protein structure 170 having the output protein sequence 150b by at least determining the conformation of the side chains of the first protein structure 170. In some cases, for each amino acid residue selected for inclusion in the output protein sequence 150b, the side chain of the first protein structure 170 may include a chemical group attached to the backbone of the first protein structure 170. In some cases, the side chain of an amino acid residue (such as the amino acid residue arginine (R) selected to occupy the first position 600a of the output protein sequence 150b) may include a backbone atom (e.g., a nitrogen (N) atom, a arginine (R ... In some cases, the side chain of an amino acid residue may include one or more atoms (e.g., side chain atoms) of the backbone of the amino acid residue. One or more atoms to which an atom is connected (e.g., side chain atoms).

[0089] In some cases, the protein structure computational model 117 may determine the side chain conformation of the first protein structure 170 based at least on the backbone conformation of the first protein structure 170. In addition, the protein structure computational model 117 may be applied to determine the side chain conformation of the first protein structure 170 with different degrees of freedom. For example, in some cases, the protein structure computational model 117 may determine the side chain conformation of each amino acid residue in at least a portion of the output protein sequence 150b by at least selecting one of a plurality of possible rotational isomers associated with the amino acid residue, each of which corresponds to a single possible conformation of atoms constituting the side chain. Determining the side chain conformation of the first protein structure 170 in this manner constrains each amino acid residue in the first protein structure 170 to one of several discrete structural variants, which may be computationally more efficient than determining the side chain conformation across continuous structural variants. Alternatively, the protein structure computational model 117 may determine the side chain conformation of each amino acid residue in at least a portion of the output protein sequence 150b by determining one or more of the following: a translation of the side chain, a rotation of the side chain, and / or a torsion angle of one or more rotatable bonds formed by side chain atoms.

[0090] FIG3B depicts a flow chart showing an example of a process 350 for hybrid protein design according to some exemplary embodiments. Referring to FIGS. 1-2 and 3A-3B, process 350 may be performed by protein design engine 110 to generate output protein sequence 150b and corresponding first protein structure 170 based at least on output protein sequence 150a. In some cases, process 350 may implement operation 206 of process 200, wherein protein structure computational model 117 selects constituent amino acid residues of the protein sequence, and determines the corresponding protein structure by determining at least the conformation of the backbone and side chains of the protein structure. In addition, in some cases, process 350 may implement operation 302 of process 300, wherein protein structure computational model 117 selects a plurality of amino acid residues that satisfy one or more criteria (such as minimization of the energy function of first protein structure 170 with output protein sequence 150b) for inclusion in output protein sequence 150b.

[0091] At 352, the protein structure computational model 117 may determine a first energy of a first protein molecule having a first plurality of amino acid residues and a first conformation. In some exemplary embodiments, the protein structure computational model 117 may generate the first protein structure 170 by at least determining an arrangement of amino acid residues for inclusion in the output protein sequence 150b and a corresponding conformation (or spatial arrangement) of constituent atoms having a minimum energy. In some cases, the protein structure computational model 117 may determine an arrangement of amino acid residues for inclusion in the output protein sequence 150b by at least selecting an amino acid residue from a corresponding set of possible amino acid residues 160 for inclusion in the output protein sequence 150b for each position in the output protein sequence 150b. Furthermore, in addition to determining the identity of the amino acid residues in each position of at least a portion of the output protein sequence 150b, the protein structure computational model 117 may also determine the conformation of the corresponding first protein structure 170, for example, by determining the backbone and side chain conformations of the first protein structure 170 having the output protein sequence 150b.

[0092] In some cases, the protein structure computational model 117 may determine the output protein sequence 150b and the first protein structure 170 formed by the output protein sequence 150b by at least generating a plurality of protein molecules having different amino acid residue sequences and / or different conformations in order to identify a specific protein sequence and conformation that satisfies one or more criteria. That is, the output protein sequence 150b and the corresponding protein structure 170 generated by the protein structure computational model 117 may satisfy one or more criteria, such as minimization of the energy of the first protein structure 170 having the output protein sequence 150b. Therefore, in some cases, after generating the first protein molecule having the first plurality of amino acid residues and the first conformation, the protein structure computational model 117 may determine the first energy of the first protein molecule, for example, by applying an energy function. In some cases, the protein structure computational model 117 may apply an energy function based on one or more of the following: ab initio quantum mechanics, density functional theory (DFT), semi-empirical methods, molecular mechanics force fields, statistical potentials, neural potentials, machine learning models (e.g., trained on structural data), etc. For example, in some cases, the energy function used to determine the first energy of the first protein molecule can be a physics-based energy function that determines a total energy for the first protein molecule that includes one or more of the following: electrostatic energy, covalent bonding energy, van der Waals energy, etc. It should be understood that different energy functions can be associated with different accuracy and computational complexity. For example, a more accurate energy function (such as an energy function based on ab initio quantum mechanics) can impose a greater computational overhead than a less accurate energy function (such as an energy function based on a molecular mechanics force field). Therefore, in some cases, the protein structure computational model 117 can apply the first energy function instead of the second energy function to determine the first energy of the first protein molecule based at least on the respective accuracy and / or computational complexity of the first energy function and the second energy function. As described in more detail below, the protein structure computational model 117 can generate additional protein molecules with different amino acid residue sequences and / or different conformations before identifying the one protein molecule with the lowest energy (e.g., total energy).

[0093] At 354, the protein structure computational model 117 may generate a second protein molecule having a second plurality of amino acid residues and a second conformation by modifying at least one of the first plurality of amino acid residues and the first conformation. In some exemplary embodiments, the protein structure computational model 117 may generate one or more additional protein molecules having a different amino acid residue sequence and / or a different conformation than the first protein molecule generated in operation 352. For example, in some cases, the protein structure computational model 117 may generate a second protein molecule having at least one of a different amino acid residue sequence and a different conformation than the first protein molecule generated in operation 352. In some cases, the second protein molecule may be generated by modifying the sequence of amino acid residues forming the first protein molecule, for example, by inserting, deleting, and / or modifying one or more of the amino acid residues in the first protein molecule. Alternatively and / or in addition, the second protein molecule may be generated by modifying the conformation of the first protein molecule, for example, by modifying one or more of the backbone conformation and the side chain conformation of the first protein molecule.

[0094] At 356, the protein structure computational model 117 may determine a second energy of the second protein molecule. In some exemplary embodiments, after generating the second protein molecule having the second plurality of amino acid residues and the second conformation, the protein structure computational model 117 may determine the second energy of the second protein molecule, for example, by applying an energy function. In some cases, the energy function used to determine the second energy of the second protein molecule may also be a physics-based energy function that determines a total energy including one or more of the following for the second protein molecule: electrostatic energy, covalent bonding energy, van der Waals energy, etc. For example, in some cases, the protein structure computational model 117 may determine the second energy of the second protein molecule by applying an energy function based on one or more of the following: ab initio quantum mechanics, density functional theory (DFT), semi-empirical methods, molecular mechanics force fields, statistical potentials, neural potentials, machine learning models (e.g., trained on structural data), etc.

[0095] At 358, the protein structure computational model 117 may generate a protein structure having the first plurality of amino acid residues and the first conformation based at least on the first energy being lower than the second energy. In some exemplary embodiments, if the first energy of the first protein molecule is lower than the second energy of the second protein molecule, the protein structure computational model 117 may generate the output protein sequence 150b of the first protein structure 170 to have the first plurality of amino acid residues forming the first protein molecule. Furthermore, in the case where the first energy of the first protein molecule is lower than the second energy of the second protein molecule, the protein structure computational model 117 may generate the first protein structure 170 to have the first conformation of the first protein molecule. Furthermore, in some cases, the protein structure computational model 117 may generate a third protein molecule having a third plurality of amino acid residues and a third conformation by modifying the first protein molecule (e.g., the first plurality of molecules and / or the first conformation) or the second protein molecule (e.g., the second plurality of molecules and / or the second conformation). In a case where the first energy of the first protein molecule is lower than the third energy of the third protein molecule, the protein structure computational model 117 may generate the first protein structure 170 to have a first conformation of the first protein molecule.

[0096] FIG4A depicts a flow chart showing another example of a process 400 for hybrid protein design according to some exemplary embodiments. Referring to FIGS. 1 to 3A, 3B, and 4A, process 400 may be performed by protein design engine 110 to verify a protein structure generated by a workflow on a computer. For example, in some cases, protein design engine 110 may perform process 400 to verify a protein structure generated by a protein structure computational model (such as protein structure computational model 117 shown in FIG1 ). Alternatively and / or in addition, protein design engine 110 may perform process 500 to verify a protein structure generated by grafting at least a portion of one protein structure (e.g., a donor protein structure) onto a corresponding portion of another protein structure (e.g., an acceptor or template protein structure).

[0097] At 402, the protein design engine 110 may apply a protein structure computational model to determine a second protein structure having at least the same protein sequence as the first protein structure based at least on the protein sequence of the first protein structure. In some exemplary embodiments, where the protein design engine 110 applies the protein design computational model 117 to determine the first protein structure 170 having the output protein sequence 150b, the protein design engine 110 may apply a different protein structure computational model to verify that the output protein sequence 150b will fold into the first protein structure 170 (e.g., having the backbone and side chain conformations of the first protein structure 170 determined by the protein design computational model 117). For example, in the example shown in FIG. 4B, the protein design engine 110 may apply a protein structure computational model 450 (which may be a different protein structure computational model than the protein structure computational model 117) to determine a second protein structure 455 formed by at least the output protein sequence 150b.

[0098] In some cases, the protein design engine 110 may apply the protein structure computational model 450 or a plurality of different protein structure computational models to generate a second protein structure 455 or a plurality of protein structures formed by the output protein sequence 150b. In addition, in some cases, the protein structure computational model 450 may be different from the protein structure computational model 117 at least because the protein structure computational model 450 includes a different machine learning model and / or implements a different structural design scheme compared to the protein structure computational model 117. For example, in some cases, where the protein structure computational model 117 determines the first protein structure 170 by at least determining the amino acid residues contained in the output protein sequence 150b and the corresponding conformations (or spatial arrangements) of the amino acid residues (e.g., the constituent atoms in each amino acid residue) , the protein structure computational model 450 may determine the second protein structure 465 based at least on the output protein sequence 150b.

[0099] At 404, the protein design engine 110 may determine / calculate a similarity metric that quantifies the difference between the first protein structure and the second protein structure. In some exemplary embodiments, the protein design engine 110 may include a structure analyzer 460 that determines a similarity metric 465 that quantifies the difference between the first protein structure 170 generated by the protein structure computational model 117 and the second protein structure 455 generated by the protein structure computational model 450. In the event that the protein design engine 110 generates multiple protein structures associated with the output protein sequence 150b, rather than just the second protein structure 465, it should be understood that the structure analyzer 460 may determine the similarity metric 465 for each protein structure. In some cases, the similarity metric 465 may include a root mean square deviation (RMSD) calculated based on the best superposition atomic coordinates of the first protein structure 170 and the second protein structure 455. However, it should be understood that the similarity metric 465 may include other values ​​that quantify the structural difference between the first protein structure 170 and the second protein structure 455. For example, in some cases, structure analyzer 460 may perform principal component analysis (PCA), in which case similarity measure 465 may include correlations between corresponding symmetric interaction matrices of first protein structure 170 and second protein structure 455. Symmetric interaction matrices of protein structures may be constructed to include relationship parameters between secondary elements such as distances, orientations, and / or other relevant structural invariants.

[0100] At 406, the protein design engine 110 may identify the protein sequence of the first protein structure as a candidate for synthesis based at least on the similarity metric satisfying one or more (pre-set) thresholds. In some exemplary embodiments, where the similarity metric 465 quantifying the structural differences between the first protein structure 170 and the second protein structure 150b satisfies the one or more thresholds, the protein design engine 110 may verify that the output protein sequence 150b will fold into the first protein structure 170 determined by the protein structure computational model 117. That is, if a different protein structure computational model (such as the protein structure computational model 450) determines the same (or sufficiently similar) protein structure for the output protein sequence 150b as the protein structure computational model 117, the protein design engine 110 may verify that the first protein structure 170 has the actual conformation of the output protein sequence 150b. In the event that the protein design engine 110 generates multiple protein structures associated with the output protein sequence 150b, rather than just the output protein structure 465, the protein design engine 110 may verify that the first protein structure 170 has the actual conformation of the output protein sequence 150b if the similarity measure 465 of a threshold amount of protein structures generated by different protein structure computational models (such as the second protein structure 455 generated by the protein structure computational model 450) satisfies the one or more thresholds.

[0101] In cases where the first protein structure 170 (e.g., a particular conformation of the first protein structure 170) is associated with one or more desired properties, if the protein design engine 110 is able to verify that the output protein sequence 150b will fold into the conformation of the first protein structure 170, the output protein sequence 150b can be identified for further analysis, including, for example, synthesis, in vitro measurements, in vivo characterization, etc. For example, the first protein structure 170 can exhibit one or more desired properties, such as expression, binding affinity for a target molecule (e.g., an antigen, such as a viral antigen or a tumor antigen), specificity for a target molecule, lack of non-specificity, stability, non-immunogenicity, human nature, absence of self-association (or non-aggregation), etc. Thus, where the protein design engine 110 is able to verify that the output protein sequence 150 b will fold into the first protein structure 170 , the output protein sequence 150 b is identified for further analysis at least because the protein design engine 110 can determine that a physical protein structure synthesized from the output protein sequence 150 b is likely to exhibit the same desired properties as the first protein structure 170 generated on the computer.

[0102] FIG5 depicts a flow chart showing another example of a process 500 for hybrid protein design according to some exemplary embodiments. Referring to FIGS. 1 to 5 , process 500 may be performed by protein design engine 110 to generate a protein structure by at least grafting a protein structure onto another protein structure. In some cases, process 200 described in FIG2 and / or process 300 described in FIG3A may be performed to generate a donor protein structure and / or an acceptor (or template) protein structure used in process 500. As described in more detail below, the donor protein structure may be a portion of a protein molecule (e.g., an antigen binding fragment (Fab), a paratope, a complementarity determining region (CDR), a variable region (Fv), etc.) that is grafted onto a corresponding portion of an acceptor (or template) protein structure to form a complete protein molecule.

[0103] At 502, the protein design engine 110 may generate a first protein structure comprising at least a first portion of a protein molecule. As noted, in some exemplary embodiments, the protein design engine 110 may apply the protein sequence computational model 113 and the protein structure computational model 117 to generate a first protein structure 170 having an output protein sequence 150b. In some cases, the first protein structure 170 generated by the protein sequence computational model 113 and the protein structure computational model 117 may be a donor structure comprising a portion of the entire protein molecule, such as a paratope, variable region (Fv), antigen binding fragment (Fab), or complementarity determining region (CDR) of an antibody. For example, in some cases, the protein design engine 110 may apply the protein sequence computational model 113 and the protein structure computational model 117 to generate a specific portion of a protein molecule (rather than the protein molecule in its entirety). In some cases, the aforementioned portion of the protein molecule may be one or more domains, each of which is an independently folded portion of the protein molecule. The first protein structure 170 used as a donor structure may include the entire portion of the protein molecule generated by the protein sequence computational model 113 and the protein structure computational model 117, or in some cases may include another sub-portion of the portion of the protein molecule. For example, in the case where the protein design engine 110 applies the protein sequence computational model 113 and the protein structure computational model 117 to generate an antigen binding fragment (Fab) of an antibody molecule, the first protein structure 170 may be one or more complementarity determining regions in the complementarity determining region (CDR) of the variable region (Fv) or the antigen binding fragment (Fab). Alternatively, the first protein structure 170 may be a portion of the entire protein molecule (e.g., antibody, etc.) generated by the protein sequence computational model 113 and the protein structure computational model 117. In the case where the protein sequence computational model 113 and the protein structure computational model 117 generate an entire protein molecule (such as an entire antibody), the first protein structure 170 can be a portion of the protein molecule identified by the protein design engine 110 for use as a donor structure, such as a paratope, variable region (Fv), antigen binding fragment (Fab), or complementarity determining region (CDR) of an antibody. In the example shown in Figures 7A to B, a donor structure comprising at least a portion of the first protein structure 170 generated by the protein sequence computational model 113 and the protein structure computational model 117 can be grafted onto a second protein structure 700 (e.g., an acceptor (or template) protein structure) to form a third protein structure 750, as depicted in the embodiment shown in Figure 7A.

[0104] At 504, the protein design engine 110 may replace the second portion of the second protein structure with the first protein structure to generate a third protein structure. In some exemplary embodiments, the protein design engine 110 may replace the second portion of the second protein structure 700 (e.g., the receptor (or template) protein structure) with the first protein structure 170 to generate the third protein structure 750. For example, in some cases, the first protein structure 170 may be a portion of a protein molecule, such as a paratope, a variable region (Fv), an antigen binding fragment (Fab), or a complementarity determining region (CDR) of an antibody. Thus, the protein design engine 110 may graft the first protein structure 170 onto the second protein structure 700 by replacing at least the corresponding portion of the second protein structure 700 with the first protein structure 170. For example, in a case where the first protein structure 170 is a first paratope, a first variable region (Fv), a first antigen binding fragment (Fab), or a first complementarity determining region (CDR) of a first antibody, the protein design engine 110 may replace a second paratope, a second variable region (Fv), a second antigen binding fragment (Fab), or a second complementarity determining region (CDR) of a second antibody with the first protein structure 170.

[0105] In some cases, the protein design engine 110 may graft the first protein structure 170 onto the second protein structure 700 to generate the third protein structure 750. In some cases, the third protein structure 750 may be a complete protein molecule (e.g., an antibody, etc.) that combines portions of multiple protein structures including the first protein structure 170 and the second protein structure 700. For example, in some cases, the third protein structure 750 may include a variable region (Fv) of the first protein structure 170 and a constant region (Fc) of the second protein structure 700. Alternatively and / or in addition, the third protein structure 750 may include an antigen binding fragment (Fab) of the first protein structure 170 and a crystallizable fragment (Fc) of the second protein structure 700. In some cases, the third protein structure 750 may include one or more complementarity determining regions (CDRs) (or hypervariable regions) of the first protein structure 170 and one or more framework regions of the second protein structure 700.

[0106] At 506, the protein design engine 110 may determine that the protein sequence of the third protein structure is generated to include at least the first portion of the first protein structure and the third portion of the second protein structure. In some exemplary embodiments, the protein design engine 110 may determine a protein sequence 760 of the third protein structure 750 that is generated to include portions of the first protein structure 170 and the second protein structure 700, each portion being associated with a different potential protein sequence. (An exemplary sequence 760 is depicted at FIG. 7B.) Thus, the protein sequence 760 of the third protein structure 750 may also include at least a portion of the output protein sequence 150a of the first protein structure 170 and the protein sequence of the second protein structure 700. For example, in some cases, the protein sequence 760 of the third protein structure 750 may include a first amino acid residue sequence that forms a variable region (Fv) of the first protein structure 170 and a second amino acid residue sequence that forms a constant region (Fc) of the second protein structure 700. Alternatively and / or in addition, the protein sequence 760 of the third protein structure 750 may include a first amino acid residue sequence that forms an antigen binding fragment (Fab) of the first protein structure 170 and a second amino acid residue sequence that forms a crystallizable fragment (Fc) of the second protein structure 700. In some cases, the protein sequence 760 of the third protein structure 750 may include a first amino acid residue sequence that forms one or more complementarity determining regions (CDRs) (or hypervariable regions) of the first protein structure 170 and a second amino acid residue sequence that forms one or more framework regions of the second protein structure 700. As described in more detail below, in some cases, the protein design engine 110 may apply a protein structure computational model to verify that the protein sequence 760 of the third protein structure 750 will assume the conformation of the third protein structure 750 formed by grafting the first protein structure 170 (e.g., a donor structure) onto the second protein structure 700 (e.g., an acceptor or template structure).

[0107] At 508, the protein design engine 110 may apply the protein structure computational model to determine a fourth protein structure having at least the same protein sequence as the third protein structure based at least on the protein sequence of the third protein structure. As noted, in some exemplary embodiments, the protein design engine 110 may validate the third protein structure 750 formed by grafting the first protein structure 170 (e.g., donor structure) onto the second protein structure 700 (e.g., acceptor or template structure) by at least validating that the protein sequence 760 of the third protein structure 750 (which includes a sequence of amino acid residues from the first protein structure 170 and the second protein structure 700) will assume the conformation of the third protein structure 750. In some cases, the protein design engine 110 may validate the third protein structure 750 by applying one or more protein structure computational models 450 to determine one or more additional protein structures 455 for evaluation against the third protein structure 750 based at least on the protein sequence 760 of the third protein structure 750. For example, in some cases, the protein design engine 110 may apply the protein structure computational model 117 used to generate the first protein structure 170 to generate the one or more additional protein structures 455 for validating the third protein structure 750. Alternatively and / or in addition, the protein design engine 110 may apply the protein structure computational model used to generate the second protein structure 700 and / or another protein structure computational model to generate the one or more additional protein structures 455 for validating the third protein structure 750. As described in more detail below, validating the third protein structure 750 may include evaluating the third protein structure 750 against the one or more additional protein structures 455 generated based on the protein sequence 760 of the third protein structure 750.

[0108] At 510, the protein design engine 110 may determine a similarity metric that quantifies the difference between the third protein structure and the fourth protein structure. In some exemplary embodiments, the structure analyzer 460 of the protein design engine 110 may determine a similarity metric 465 that quantifies the difference between the third protein structure 750 and the one or more additional protein structures 455 generated based on the protein sequence 460 of the third protein structure 750. In some cases, the similarity metric 465 may include a root mean square deviation (RMSD) that quantifies the structural difference between the third protein structure 750 and each of the additional protein structures 455 generated based on the protein sequence 760 of the third protein structure 750. Further, in some cases, the similarity metric 465 may indicate a likelihood that the protein sequence 760 of the third protein structure 750 will assume the conformation of the third protein structure 750 generated by grafting the first protein structure 170 onto the second protein structure 700.

[0109] At 512, the protein design engine 110 may identify the protein sequence of the third protein structure as a candidate for synthesis based at least on the similarity metric satisfying one or more thresholds. In some exemplary embodiments, where the similarity metric 465 quantifying the structural differences between the third protein structure 750 and the one or more additional protein structures 455 satisfies one or more thresholds, the protein design engine 110 may verify that the protein sequence 760 will fold into the third protein structure 750 generated by grafting the first protein structure 170 onto the second protein structure 700. In some cases, if the one or more protein structure computational models 450 determine a protein structure for the protein sequence 760 that is the same (or sufficiently similar) as the first protein structure 170 grafted onto the second protein structure 700, the protein design engine 110 may verify that the third protein structure 750 exhibits the actual conformation of the protein sequence 760. In the case where the protein design engine 110 generates multiple protein structures 455 having a protein sequence 760 of a third protein structure 750, the protein design engine 110 may verify that the third protein structure 750 has an actual conformation of the protein sequence 760 if the similarity metrics 465 of a threshold amount of the additional protein structures 455 satisfy the one or more thresholds.

[0110] In some exemplary embodiments, upon verification of the third protein structure 750, the protein design engine 110 may select the corresponding protein sequence 760 for further analysis, including, for example, synthesis, in vitro measurement, in vivo characterization, etc. For example, in some cases, the third protein structure 760 or one or more of its constituent components (such as the first protein structure 170 and / or the second protein structure 700) may exhibit one or more desired properties, such as expression, binding affinity for a target molecule (e.g., an antigen, such as a viral antigen or a tumor antigen), specificity for a target molecule, lack of non-specificity, stability, non-immunogenicity, human nature, absence of self-association (or non-aggregation), etc. Thus, in the event that the protein design engine 110 is able to verify that the protein sequence 760 will fold into the third protein structure 750, the protein design engine 110 may identify the protein sequence 760 of the third protein structure 750 for further analysis at least because the protein design engine 110 may determine that a physical protein structure synthesized from the protein sequence 760 is likely to exhibit the same desired properties as the third protein structure 760. For example, where the first protein structure 170 is a paratope, antigen binding fragment (Fab), variable region (Fv), or complementarity determining region (CDR) that exhibits binding affinity for a target molecule and the third protein structure 750 is generated to include the first protein structure 170 grafted onto the second protein structure 700 as the corresponding constant region, crystallizable fragment (Fc), or framework region of the third protein structure 750, the protein design engine 110 may determine that a protein structure synthesized from the protein sequence 760 of the third protein structure 750 is likely to exhibit the same binding affinity for the target molecule.

[0111] FIG8A depicts a schematic diagram showing an example of a process for hybrid protein design according to some exemplary embodiments, wherein an alternate protein sequence is generated for an antigen binding fragment of the antibody trastuzumab. Referring to FIG8A , the protein design engine 110 may determine one or more interface residues based at least on a co-crystal structure 800 of a first protein structure 825 (corresponding to the antibody trastuzumab) combined with a second protein structure 850 (corresponding to the protein human epidermal growth factor receptor 2 (HER2)). In this context, an interface residue may refer to an amino acid residue in a protein molecule (such as a first protein structure 825 of an antibody molecule trastuzumab) that interacts and contacts another protein molecule (such as a second protein structure 850 of a human epidermal growth factor receptor 2 (HER2) protein molecule). As shown in FIG8A , the interface residues of the antibody trastuzumab may include constituent amino acid residues of an antigen binding fragment (Fab) of the antibody trastuzumab. In some cases, the interface residues of the antibody trastuzumab may include amino acid residues that form a light chain (V) of the antibody trastuzumab. L ) and heavy chain The amino acid residues of the variable domain.

[0112] Referring again to FIG. 8A , in some exemplary embodiments, one or more alternate protein sequences for the antibody trastuzumab (or some portion thereof) may be generated by at least applying a protein sequence computational model 113. In the example shown in FIG. 8A , the protein sequence computational model 113 may be an autoencoder (e.g., a variational autoencoder, etc.) that generates each alternate protein sequence for the antibody trastuzumab by sampling the data distribution occupied by the encoding of the known protein sequence. Thus, for each sampling iteration of the data distribution, the protein sequence computational model 113 may encode at least a portion of the original protein sequence of the antibody trastuzumab, and then decode an intermediate sequence having at least one of a disruption (e.g., insertion, deletion, and / or modification of amino acid residues) and a length change relative to the protein sequence of trastuzumab.

[0113] In some cases, the protein design engine 110 may apply the protein sequence computational model 113 to generate an alternate sequence for a portion of the antibody trastuzumab (e.g., a paratope, an antigen binding fragment (Fab), a variable region (Fv), a complementarity determining region (CDR), etc.). Alternatively, the protein design engine 110 may apply the protein sequence computational model 113 to generate an alternate sequence for the entire trastuzumab molecule. For example, in some cases, the protein design engine 110 may apply the protein sequence computational model 113 to generate a plurality of proposed protein sequences based on the original protein sequence of the antibody trastuzumab (or a portion thereof), and then determine possible sets of amino acid residues for each position in the alternate trastuzumab protein sequence based on the proposed protein sequence. In some cases, the possible sets of amino acid residues for a position in the alternate trastuzumab protein sequence may include one or more amino acid residues that appear in the position with sufficient frequency across the proposed protein sequences generated by the protein sequence computational model 113.

[0114] In some exemplary embodiments, an alternate protein sequence for the antibody trastuzumab (or a portion thereof) may be generated by applying a protein structure computational model 117 that determines an alternate trastuzumab sequence based on possible amino acid residue groups for each position and a corresponding conformation of a protein molecule having the alternate protein sequence. In cases where the conformation presented by the alternate protein sequence is sufficiently similar to the conformation of the original trastuzumab molecule (e.g., the first protein structure 825 in the co-crystal structure 800), the protein design engine 110 may identify the alternate protein sequence for further analysis, including, for example, synthesis, in vitro measurements, in vivo characterization, etc. For example, the protein design engine 110 may apply a protein structure computational model (which may or may not be the protein structure computational model 117 that was applied to generate the alternate sequence and structure of trastuzumab) to determine the conformation of the protein molecule having the alternate protein sequence. In the event that the conformation of the protein molecule with the alternate protein sequence exhibits sufficient similarity to the original structure of trastuzumab, such as when a similarity metric (e.g., root mean square deviation (RMSD), etc.) that quantifies the difference between the two molecules satisfies one or more thresholds, the protein design engine 110 may identify the alternate sequence for trastuzumab for further analysis (e.g., synthesis, in vitro measurements, in vivo characterization, etc.).

[0115] FIG8B depicts a violin chart 860 showing predicted binding affinities of alternate trastuzumab sequences generated by different hybrid protein design workflows and a violin chart 865 showing measured affinities, according to some exemplary embodiments. Referring to FIG8B , the protein structures generated by each hybrid protein design workflow can be evaluated by calculating an interface energy (dG separation) that indicates the predicted binding affinity of each alternate trastuzumab protein sequence (to human epidermal growth factor receptor 2 (HER2)) by quantifying at least the difference between a first energy of the bound trastuzumab human epidermal growth factor receptor 2 (HER2) protein complex and a second energy of the separated protein molecules. The predicted binding affinity of the alternate trastuzumab protein sequence (to human epidermal growth factor receptor 2 (HER2)) can also be determined by measuring, for example, the dissociation constant (K D ) to empirically determine the binding affinity of each alternating trastuzumab protein sequence. Violin graph 860 and violin graph 865 show the distribution of interface energy (dG separation) and measured binding affinity (pkd) across a first population of non-binding alternating trastuzumab protein sequences (labeled as negative) and a second population of binding alternating trastuzumab protein sequences, respectively. The two alternating trastuzumab protein sequence populations were generated by a hybrid protein design workflow that includes a protein sequence computational model trained based on a training data set that includes a first plurality of labeled non-binding protein sequences and a second plurality of labeled binding protein sequences. As shown in graphs 850 and 860, the majority of the binding alternating trastuzumab protein sequences generated by the hybrid protein design workflow exhibit high binding affinity, as indicated by higher dissociation constants and lower interface energies.

[0116] FIG8C depicts a graph 870 showing the stability of each alternate trastuzumab protein structure as indicated by the total energy of the molecule. Graph 870 in FIG8C shows that alternate trastuzumab protein sequences generated by various hybrid protein design workflows described herein fold into more stable three-dimensional structures (e.g., having lower total energy) than conventional structurally agnostic protein sequence design techniques.

[0117] In view of the above specific implementation of the subject matter, the present application discloses the following list of examples, wherein one feature of a single example or a combination of more than one feature of the example, and optionally a combination with one or more features of one or more other examples, are other examples that also fall within the disclosure scope of the present application:

[0118] Item 1: A computer-implemented method comprising: identifying a protein sequence computational model and a protein structure computational model; applying the protein sequence computational model to generate a plurality of proposed protein sequences based at least on an input protein sequence; identifying a set of possible amino acid residues for each position in at least a portion of an output protein sequence based at least on the plurality of proposed protein sequences; and using the protein structure computational model to generate a first protein structure having the output protein sequence by: applying the protein structure computational model to select a possible amino acid residue from the set of possible amino acid residues for inclusion in the output protein sequence for each position in at least the portion of the output protein sequence.

[0119] Item 2: The method according to Item 1, further comprising: aligning the multiple proposed protein sequences to generate multiple aligned protein sequences; and identifying the set of possible amino acid residues for each position in at least the portion of the output protein sequence based at least on the multiple aligned protein sequences.

[0120] Item 3: A method according to any one of items 1 to 2, wherein the multiple proposed protein sequences are aligned by applying one or more of the following: dynamic programming, progressive alignment, hierarchical alignment, iterative alignment, motif discovery, deep learning models and hidden Markov models.

[0121] Item 4: A method according to any one of Items 1 to 4, wherein the identifying the set of possible amino acid residues for each position in at least the portion of the output protein sequence comprises: identifying a first amino acid residue rather than a second amino acid residue for inclusion in the set of possible amino acid residues.

[0122] Item 5: A method according to Item 4, wherein the identifying of the set of possible amino acid residues for each position in at least the portion of the output protein sequence further comprises: determining a first frequency of a first amino acid residue appearing at the position across the multiple proposed protein sequences generated by the protein sequence computational model, determining a second frequency of a second amino acid residue appearing at the position across the multiple proposed protein sequences generated by the protein sequence computational model, and identifying the first amino acid residue rather than the second amino acid residue for inclusion in the set of possible amino acid residues for the position based at least on the first frequency and the second frequency.

[0123] Item 6: A method according to Item 5, wherein the first amino acid residue is identified for inclusion in the set of amino acid residues at least based on the first frequency satisfying one or more thresholds, and wherein the second amino acid residue is identified for exclusion from the set of possible amino acid residues at least based on the second frequency failing to satisfy the one or more thresholds.

[0124] Item 7: A method according to Item 6, wherein the identifying of the set of possible amino acid residues for the position in the output protein sequence further comprises determining the one or more thresholds based on at least one of: the maximum, minimum, median, mean and mode of the frequency of each amino acid residue in a plurality of amino acid residues appearing at the position across the plurality of proposed protein sequences generated by the protein sequence computational model.

[0125] Item 8: A method according to any one of items 1 to 7, wherein the set of possible amino acid residues includes some but not all of the following: alanine, arginine, asparagine, aspartic acid, cysteine, glutamic acid, glutamine, glycine, histidine, isoleucine, leucine, lysine, methionine, phenylalanine, proline, serine, threonine, tryptophan, tyrosine, valine, selenocysteine ​​and pyrrolysine.

[0126] Item 9: A method according to any one of Items 1 to 8, wherein the protein structure computational model generates the first protein structure by at least: determining the identity and conformation of the amino acid residues occupying each position in at least the portion of the output protein sequence based at least on the energy of the first protein structure having the output protein sequence.

[0127] Item 10: A method according to Item 9, wherein the protein structure computational model determines the identity and the conformation of the amino acid residue occupying each position in at least the portion of the output protein sequence by at least the following: modifying at least one of the identity and the conformation of the amino acid residue to minimize the energy of the first protein structure.

[0128] Item 11: A method according to Item 10, wherein the computational model of the protein structure modifies at least one of the identity and conformation of the amino acid residue occupying the position by at least: (i) changing the conformation of the amino acid residue occupying the position, or (ii) selecting different possible amino acid residues for the position from a group of amino acid residues associated with the position.

[0129] Item 12: A method according to any one of items 9 to 11, wherein the protein structure computational model determines the identity and the conformation of the amino acid residue occupying each position in at least the portion of the output protein sequence by at least the following: determining a first energy of the first protein structure having a first possible amino acid residue from the set of possible amino acid residues, determining a second energy of the first protein structure having a second possible amino acid residue from the set of possible amino acid residues, and generating the first protein structure to include the first possible amino acid residue instead of the second possible amino acid residue based at least on the first energy being lower than the second energy.

[0130] Item 13: A method according to Item 12, wherein the protein structure computational model further determines the identity and the conformation of the amino acid residue occupying each position in at least the portion of the output protein sequence by at least the following: determining a third energy of the first protein structure having a first conformation of the first possible amino acid residue, determining a fourth energy of the first protein structure having a second conformation of the first possible amino acid residue, and generating the first protein structure as containing the first conformation of the first possible amino acid residue instead of the second conformation of the first possible amino acid residue based at least on the third energy being lower than the fourth energy.

[0131] Item 14: The method according to any one of Items 1 to 14, wherein the protein structure computational model generates the first protein structure by at least: determining a first backbone conformation of a first skeleton of the first protein structure having the output protein sequence.

[0132] Item 15: The method according to Item 14, wherein the first backbone of the first protein structure is a continuous atomic chain formed by connecting multiple backbone atoms from each amino acid residue in the output protein sequence.

[0133] Item 16: The method according to Item 15, wherein the plurality of skeleton atoms are nitrogen (N) atoms, Atoms and the atomic sequence of carboxyl carbon (C) atoms.

[0134] Item 17: A method according to any one of Items 14 to 16, wherein determining the first backbone conformation of the first backbone of the first protein structure comprises determining that the first backbone of the first protein structure has the same conformation as at least a portion of the second backbone of the second protein structure.

[0135] Item 18: The method of Item 17, wherein the second protein structure is associated with a protein sequence determined to have one or more desired properties.

[0136] Item 19: The method according to Item 18, wherein the protein sequence having the one or more desired properties is the input protein sequence or another protein sequence.

[0137] Item 20: The method according to any one of Items 17 to 19, wherein the second backbone of the second protein structure exhibits the second backbone conformation of the second protein structure in an unbound state or the third backbone conformation of the second protein structure bound to a target molecule.

[0138] Item 21: A method according to any one of Items 14 to 20, wherein determining the first backbone conformation of the first backbone of the first protein structure comprises: determining the translation of the multiple backbone atoms contained in each amino acid residue in at least a portion of the output protein sequence.

[0139] Item 22: A method according to any one of Items 14 to 21, wherein determining the first backbone conformation of the first protein structure comprises: determining the rotation of the plurality of backbone atoms contained in each amino acid residue in at least a portion of the output protein sequence.

[0140] Item 23: A method according to any one of Items 14 to 22, wherein determining the first backbone conformation of the first protein structure comprises: determining the torsion angles of one or more rotatable bonds formed by the multiple backbone atoms contained in each amino acid residue in the output protein sequence.

[0141] Item 24: A method according to any one of Items 14 to 23, wherein the protein structure computational model further generates the first protein structure as comprising, at each position in at least a portion of the first skeleton having the first backbone conformation, a plurality of side chain atoms selected for inclusion in the amino acid residue at the corresponding position in the output protein sequence.

[0142] Item 25: The method according to Item 24 further includes: applying different protein structure computational models to determine a second protein structure having at least the output protein sequence based at least on the output protein sequence; and determining a similarity measure that quantifies the difference between the second protein structure and the first protein structure generated to have the first backbone conformation.

[0143] Item 26: The method of Item 25, further comprising: identifying the output protein sequence as a candidate for synthesis based at least on the similarity measure satisfying one or more thresholds.

[0144] Item 27: A method according to any one of Items 14 to 26, wherein the protein structure computational model further generates the first protein structure by at least the following: determining, for each position in at least a portion of the first skeleton having the first backbone conformation, the side chain conformation of the amino acid residue selected for inclusion at the corresponding position in the output protein sequence.

[0145] Item 28: The method according to Item 27, wherein the side chain conformation of the amino acid residue is determined based at least on the first backbone conformation of the first framework of the first protein structure.

[0146] Item 29: The method according to any one of Items 27 to 28, wherein said determining the side chain conformation of the amino acid residue selected for inclusion in the output protein sequence comprises: selecting a rotamer from a plurality of possible rotamers.

[0147] Item 30: The method according to Item 29, wherein each rotamer of the plurality of possible rotamers comprises a different combination of torsion angles of one or more rotatable bonds formed by a plurality of side chain atoms in the amino acid residue.

[0148] Item 31: A method according to any one of Items 27 to 30, wherein determining the side chain conformation of the amino acid residue selected for inclusion in the output protein sequence comprises: determining the torsion angle of one or more rotatable bonds formed by multiple side chain atoms in the amino acid residue.

[0149] Item 32: The method according to any one of Items 27 to 31, wherein said determining the side chain conformation of the amino acid residue selected for inclusion in the output protein sequence comprises: determining the translation of a plurality of side chain atoms in the amino acid residue.

[0150] Item 33: The method according to any one of Items 27 to 32, wherein said determining the side chain conformation of the amino acid residue selected for inclusion in the output protein sequence comprises: determining the rotation of a plurality of side chain atoms in the amino acid residue.

[0151] Item 34: A method according to any one of items 1 to 33, further comprising: applying a property analysis model to determine the properties of each protein sequence included in the multiple proposed protein sequences; identifying at least one protein sequence among the multiple proposed protein sequences for exclusion based at least on the properties of each protein sequence; and excluding the at least one protein sequence from the multiple proposed protein sequences, and then identifying the set of possible amino acid residues for each position in the output protein sequence based at least on the remaining multiple proposed protein sequences.

[0152] Item 35: A method according to any one of items 1 to 34, further comprising: identifying a first portion of a second protein structure; and generating a third protein structure by replacing the first portion of the second protein structure with at least a second portion of the first protein structure.

[0153] Item 36: The method according to Item 35 further includes: determining a third protein sequence of the third protein structure generated to include the second part of the first protein structure and the third part of the second protein structure; applying the protein structure computational model and / or a different protein structure computational model to determine a fourth protein structure having at least the third protein sequence based at least on the third protein sequence; determining a similarity metric that quantifies the difference between the third protein structure and the fourth protein structure; and identifying the third protein sequence as a candidate for synthesis based at least on the similarity metric satisfying one or more thresholds.

[0154] Item 37: The method according to any one of Items 35 to 36, wherein the second protein structure is selected based at least on the fact that the second protein structure exhibits one or more desired properties.

[0155] Item 38: A method according to any one of Items 35 to 37, wherein the first portion of the second protein structure comprises a first antigen binding site of a first antibody having the second protein structure, and wherein the second portion of the first protein structure comprises a second antigen binding site of a second antibody having the first protein structure.

[0156] Item 39: A method according to any one of Items 35 to 38, wherein the first portion of the second protein structure comprises a first paratope of a first antibody having the second protein structure, and wherein the second portion of the first protein structure comprises a second paratope of a second antibody having the first protein structure.

[0157] Item 40: A method according to any one of Items 35 to 39, wherein the first portion of the second protein structure comprises a first complementarity determining region (CDR) of a first antibody having the second protein structure, and wherein the second portion of the first protein structure comprises a second complementarity determining region (CDR) of a second antibody having the first protein structure.

[0158] Item 41: A method according to any one of Items 1 to 40, wherein the protein structure computational model generates the first protein structure by at least: determining a plurality of amino acid residues for inclusion in the output protein sequence and corresponding conformations of the plurality of amino acid residues based at least on one or more criteria.

[0159] Item 42: The method of Item 41, wherein the one or more criteria comprise minimizing an energy function of the first protein structure having the plurality of amino acid residues selected for inclusion in the output protein sequence.

[0160] Item 43: A method according to any one of items 41 to 42, wherein the protein structure computational model generates the first protein structure by at least the following: determining a first energy of a first protein molecule having a first plurality of amino acid residues and a first conformation; generating a second protein molecule having a second plurality of amino acid residues and a second conformation by modifying at least one of the first plurality of amino acid residues and the first conformation; determining a second energy of the second protein molecule; and generating the first protein structure to have the first plurality of amino acid residues and the first conformation at least based on the first energy being lower than the second energy.

[0161] Item 44: A method according to any one of Items 1 to 43, wherein the protein sequence computational model comprises one or more machine learning models trained to generate the plurality of proposed protein sequences based on the input protein sequence.

[0162] Item 45: A method according to any one of items 1 to 44, wherein the protein sequence computational model generates the multiple proposed protein sequences by at least: sampling a data distribution populated by multiple encoded protein sequences, and wherein each sampling of the data distribution generates an encoding of a protein sequence having at least one of a disruption and a length change relative to the input protein sequence.

[0163] Item 46: A system comprising: at least one data processor; and at least one memory storing instructions which, when executed by the at least one data processor, result in operations including those of the method of any one of items 1 to 45.

[0164] Item 47: A non-transitory computer readable medium storing instructions which, when executed by at least one data processor, result in operations including the method according to any one of items 1 to 45.

[0165] 9 depicts a block diagram showing an example of a computing system 900 according to some exemplary embodiments. Referring to FIG1-9, computing system 900 may be used to implement protein design engine 110, molecular analysis engine 120, client device 130, and / or any components thereof.

[0166] As shown in FIG9 , the computing system 900 may include a processor 910, a memory 920, a storage device 930, and an input / output device 940. The processor 910, the memory 920, the storage device 930, and the input / output device 940 may be interconnected via a system bus 950. The processor 910 is capable of processing instructions for execution within the computing system 900. Such executed instructions may implement, for example, one or more components of the protein design engine 110, the analysis engine 120, the client device 130, etc. In some exemplary embodiments, the processor 910 may be a single-threaded processor. Alternatively, the processor 910 may be a multi-threaded processor. The processor 910 is capable of processing instructions stored on the memory 920 and / or the storage device 930 to display graphical information for a user interface provided via the input / output device 940.

[0167] The memory 920 is a computer-readable medium, such as a volatile or non-volatile computer-readable medium, that stores information within the computing system 900. For example, the memory 920 may store a data structure representing a configuration object database. The storage device 930 can provide persistent storage for the computing system 900. The storage device 930 may be a floppy disk device, a hard disk device, an optical disk device, a tape device, or other suitable persistent storage device. The input / output device 940 provides input / output operations for the computing system 900. In some exemplary embodiments, the input / output device 940 includes a keyboard and / or a pointing device. In various specific implementations, the input / output device 940 includes a display unit for displaying a graphical user interface.

[0168] According to some exemplary embodiments, the input / output device 940 may provide input / output operations for network devices. For example, the input / output device 940 may include an Ethernet port or other networking port to communicate with one or more wired and / or wireless networks (e.g., a local area network (LAN), a wide area network (WAN), the Internet).

[0169] In some exemplary embodiments, the computing system 900 can be used to execute various interactive computer software applications that can be used to organize, analyze and / or store data in various formats. Alternatively, the computing system 900 can be used to execute any type of software application. These applications can be used to perform various functions, such as planning functions (e.g., generating, managing, editing electronic spreadsheet documents, word processing documents and / or any other objects, etc.), computing functions, communication functions, etc. The application can include various additional functions or can be independent computing products and / or functions. After activation within the application, the function can be used to generate a user interface provided via the input / output device 940. The user interface can be generated by the computing system 900 and presented to the user (e.g., on a computer screen monitor, etc.).

[0170] One or more aspects or features of the subject matter described herein can be implemented with digital electronic circuits, integrated circuits, specially designed ASICs, field programmable gate arrays (FPGAs), computer hardware, firmware, software, and / or combinations thereof. These various aspects or features can include a specific implementation in one or more computer programs, which are executable and / or interpretable on a programmable system, which includes at least one programmable processor (which can be dedicated or general, coupled to receive data and instructions from it and send data and instructions to it), a storage system, at least one input device, and at least one output device. A programmable system or computing system can include a client and a server. Typically, the client and the server are remotely arranged from each other, and generally interact through a communication network. The relationship between the client and the server is generated by means of a computer program running on each computer and a client-server relationship between each other.

[0171] These computer programs may also be referred to as programs, software, software applications, applications, components or codes, including machine instructions for programmable processors, and may be implemented in high-level procedural and / or object-oriented programming languages ​​and / or in assembly / machine languages. As used herein, the term "machine-readable medium" refers to any computer product, device and / or device (such as, for example, a disk, an optical disk, a memory and a programmable logic device (PLD)) for providing machine instructions and / or data to a programmable processor, including a machine-readable medium that receives machine instructions as a machine-readable signal. The term "machine-readable signal" refers to any signal for providing machine instructions and / or data to a programmable processor. A machine-readable medium may store such machine instructions non-temporarily (such as, for example, a non-temporary solid-state memory or a magnetic hard drive or any equivalent storage medium). A machine-readable medium may store such machine instructions in a temporary manner (such as, for example, a processor cache or other random access memory associated with one or more physical processor cores) alternatively or additionally.

[0172] To provide interaction with a user, one or more aspects or features of the subject matter described herein may be implemented on a computer having a display device (such as, for example, a cathode ray tube (CRT) or a liquid crystal display (LCD) or a light emitting diode (LED) monitor for displaying information to a user) and a keyboard and pointing device (such as, for example, a mouse or trackball, through which a user can provide input to the computer). Other kinds of devices may also be used to provide interaction with a user. For example, the feedback provided to the user may be any form of sensory feedback, such as, for example, visual feedback, auditory feedback, or tactile feedback; input from the user may be received in any form, including sound, voice, or tactile input. Other possible input devices include touch screens or other touch-sensitive devices, such as single-point or multi-point resistive or capacitive tracking pads, voice recognition hardware and software, optical scanners, optical pointers, digital image capture devices and associated interpretation software, and the like.

[0173] In the above description and claims, phrases such as "at least one" or "one or more" may appear, followed by a list of combinations of elements or features. The term "and / or" may also appear in a list of two or more elements or features. Unless otherwise implicitly or explicitly contradicted by the context in which it is used, the phrase is intended to mean any element or feature listed alone, or any other described element or feature combined with any other described element or feature. For example, the phrases "at least one of A and B"; "one or more of A and B"; "A and / or B" are each intended to mean "single A, single B, or A and B together". A similar interpretation also applies to lists including three or more items. For example, the phrases "at least one of A, B, and C"; "one or more of A, B, and C"; and "A, B, and / or C" are each intended to mean "single A, single B, single C, A and B together, A and C together, B and C together, or A and B and C together". Use of the term "based on" above and in the claims is intended to mean "based, at least in part, on" such that non-recited features or elements are also permissible.

[0174] Depending on the desired configuration, the subject matter described herein may be embodied in systems, devices, methods and / or articles. The embodiments described in the foregoing description do not represent all embodiments consistent with the subject matter described herein. Instead, they are only some examples consistent with aspects related to the described subject matter. Although some variations have been described in detail above, other modifications or additions are possible. In particular, in addition to those features and / or variations described herein, other features and / or variations may also be provided. For example, the above-mentioned specific implementations may be directed to various combinations and sub-combinations of the disclosed features and / or to combinations and sub-combinations of several further features disclosed above. In addition, the logical flows depicted in the accompanying drawings and / or described herein do not necessarily require the specific order shown or the sequential order to achieve the desired results. Other specific implementations may be within the scope of the following claims.

Claims

1. A computer-implemented method comprising: Identify computational models of protein sequences and computational models of protein structures; applying the protein sequence computational model to generate a plurality of proposed protein sequences based at least on an input protein sequence; identifying a set of possible amino acid residues for each position in at least a portion of an output protein sequence based at least on the plurality of proposed protein sequences; as well as A first protein structure having the output protein sequence is generated using the protein structure computational model by applying the protein structure computational model to select, for each position in at least the portion of the output protein sequence, a possible amino acid residue from the set of possible amino acid residues for inclusion in the output protein sequence.

2. The method according to claim 1, further comprising: aligning the plurality of proposed protein sequences to generate an aligned plurality of protein sequences; as well as The set of possible amino acid residues for each position in at least the portion of the output protein sequence is identified based at least on the aligned plurality of protein sequences.

3. The method according to any one of claims 1 to 2, wherein the plurality of proposed protein sequences are aligned by applying one or more of the following: dynamic programming, progressive alignment, hierarchical alignment, iterative alignment, motif discovery, deep learning models, and hidden Markov models.

4. The method according to any one of claims 1 to 3, wherein identifying the set of possible amino acid residues for each position in at least the portion of the output protein sequence comprises: The first amino acid residue, but not the second amino acid residue, is identified for inclusion in the set of possible amino acid residues.

5. The method of claim 4, wherein identifying the set of possible amino acid residues for each position in at least the portion of the output protein sequence further comprises determining a first frequency at which a first amino acid residue occurs at the position across the plurality of proposed protein sequences generated by the protein sequence computational model, determining a second frequency at which a second amino acid residue occurs at the position across the plurality of proposed protein sequences generated by the protein sequence computational model, and The first amino acid residue is identified, but not the second amino acid residue, for inclusion in the set of possible amino acid residues for the position based on at least the first frequency and the second frequency.

6. The method of claim 5 , wherein the first amino acid residue is identified for inclusion in a set of amino acid residues based at least on the first frequency satisfying one or more thresholds, and wherein the second amino acid residue is identified for exclusion from the set of possible amino acid residues based at least on the second frequency failing to satisfy the one or more thresholds.

7. The method of claim 6, wherein identifying the set of possible amino acid residues for the position in the output protein sequence further comprises The one or more thresholds are determined based on at least one of: a maximum, a minimum, a median, a mean, and a mode of frequencies at which each amino acid residue in a plurality of amino acid residues appears at the position across the plurality of proposed protein sequences generated by the protein sequence computational model.

8. A method according to any one of claims 1 to 7, wherein the set of possible amino acid residues includes some but not all of the following: alanine, arginine, asparagine, aspartic acid, cysteine, glutamic acid, glutamine, glycine, histidine, isoleucine, leucine, lysine, methionine, phenylalanine, proline, serine, threonine, tryptophan, tyrosine, valine, selenocysteine ​​and pyrrolysine.

9. A method according to any one of claims 1 to 8, wherein the protein structure computational model generates the first protein structure by at least: determining the identity and conformation of the amino acid residues occupying each position in at least the portion of the output protein sequence based at least on the energy of the first protein structure having the output protein sequence.

10. The method of claim 9, wherein the protein structure computational model determines the identity and the conformation of the amino acid residue occupying each position in at least the portion of the output protein sequence by at least: modifying at least one of the identity and the conformation of the amino acid residue to minimize the energy of the first protein structure.

11. The method of claim 10, wherein the computational model of protein structure modifies at least one of the identity and conformation of the amino acid residue occupying the position by at least: (i) changing the conformation of the amino acid residue occupying the position, or (ii) selecting a different possible amino acid residue for the position from a set of amino acid residues associated with the position.

12. The method according to any one of claims 9 to 11, wherein the protein structure computational model determines the identity and the conformation of the amino acid residue occupying each position in at least the portion of the output protein sequence by at least: determining a first energy of the first protein structure having a first possible amino acid residue from the set of possible amino acid residues, determining a second energy of the first protein structure having a second possible amino acid residue from the set of possible amino acid residues, and The first protein structure is generated to include the first possible amino acid residue instead of the second possible amino acid residue based at least on the first energy being lower than the second energy.

13. The method of claim 12, wherein the protein structure computational model further determines the identity and the conformation of the amino acid residue occupying each position in at least the portion of the output protein sequence by at least: determining a third energy of the first protein structure having the first conformation of the first possible amino acid residue, determining a fourth energy of the first protein structure having a second conformation of the first possible amino acid residue, and The first protein structure is generated to include the first conformation of the first possible amino acid residue instead of the second conformation of the first possible amino acid residue based at least on the third energy being lower than the fourth energy.

14. The method according to any one of claims 1 to 13, further comprising: applying a property analysis model to determine a property of each protein sequence included in the plurality of proposed protein sequences; identifying at least one protein sequence in the plurality of proposed protein sequences for exclusion based at least on the characteristic of each protein sequence; as well as The at least one protein sequence is excluded from the plurality of proposed protein sequences, and thereafter the set of possible amino acid residues for each position in the output protein sequence is identified based at least on the remaining plurality of proposed protein sequences.

15. The method according to any one of claims 1 to 14, further comprising: identifying a first portion of a second protein structure; as well as A third protein structure is generated by replacing at least the first portion of the second protein structure with at least the second portion of the first protein structure.

16. The method according to claim 15, further comprising: determining a third protein sequence generated as the third protein structure comprising the second portion of the first protein structure and the third portion of the second protein structure; applying the protein structure computational model and / or a different protein structure computational model to determine at least a fourth protein structure having the third protein sequence based at least on the third protein sequence; determining a similarity metric that quantifies differences between the third protein structure and the fourth protein structure; as well as The third protein sequence is identified as a candidate for synthesis based at least on the similarity measure satisfying one or more thresholds.

17. The method of any one of claims 15 to 16, wherein the second protein structure is selected based at least on the fact that the second protein structure exhibits one or more desired properties.

18. The method of any one of claims 15 to 17, wherein the first portion of the second protein structure comprises a first antigen binding site of a first antibody having the second protein structure, and wherein the second portion of the first protein structure comprises a second antigen binding site of a second antibody having the first protein structure.

19. The method of any one of claims 15 to 18, wherein the first portion of the second protein structure comprises a first paratope of a first antibody having the second protein structure, and wherein the second portion of the first protein structure comprises a second paratope of a second antibody having the first protein structure.

20. The method of any one of claims 15 to 19, wherein the first portion of the second protein structure comprises a first complementarity determining region (CDR) of a first antibody having the second protein structure, and wherein the second portion of the first protein structure comprises a second complementarity determining region (CDR) of a second antibody having the first protein structure.

21. A system comprising: at least one data processor; as well as At least one memory storing instructions which, when executed by the at least one data processor, result in operations including the method according to any one of claims 1 to 20.

22. A non-transitory computer readable medium storing instructions which, when executed by at least one data processor, result in operations including the method of any one of claims 1 to 20.

Images