Determining the feasibility of peptide production

Abstract

An approach is provided for predicting the manufacturability of peptides. Requests for information related to the manufacturability of peptides can be received. The determination of whether a peptide is predicted to be synthesizable can be made, for example, by using a machine learning model. The machine learning model can be trained with data including manufacturer specifications, descriptions related to the peptide, and peptide characteristics. A second determination can be made regarding whether a peptide is predicted to be soluble, using the same or a different machine learning model trained with peptide solubility data. If a peptide is predicted to be soluble and synthesizable, a manufacturability score for the peptide can be determined. The manufacturability score may correspond to or indicate the likelihood of successfully manufacturing the peptide.

Description

Background Art

[0001] Cross - reference to Related Applications This application claims the benefit of U.S. Provisional Patent Application No. 63 / 505,231, filed May 31, 2023, the entire contents of which are incorporated herein by reference in their entirety for all purposes.

[0002] Peptides are short chains of amino acids that can be used for various biological purposes. Peptides can be artificially manufactured for multiple applications, such as for use in drug discovery or supplements. Depending on the details of the manufacturing process, factors related to the length and sequence of the peptide can prevent a significant percentage of the peptides from being synthesized properly. Other factors, such as pH, temperature, and storage conditions, can all affect the stability of the peptide in solution. In order to make the manufactured peptide usable, the peptide must be synthesized correctly and solubilized. Therefore, in order to minimize manufacturing failures, there is a need for a method to accurately predict whether the production of a peptide can be carried out successfully.

Summary of the Invention

Means for Solving the Problems

[0003] Various embodiments according to the present disclosure will be described with reference to the drawings.

Brief Description of the Drawings

[0004] [Figure 1] Illustrative methods that can be used according to various embodiments are shown.

[0005] [Figure 2] Illustrative representations of manufacturability analysis systems that can be utilized according to various embodiments are shown. \

[0006] [Figure 3] A and B show exemplary histograms of predicted manufacturability according to various embodiments.

[0007] [Figure 4] Exemplary receiver operating characteristic curves for various embodiments are shown for predicted manufacturability.

[0008] [Figure 5] Exemplary calibration curves for converting manufacturability scores into manufacturability probabilities are shown for various embodiments.

[0009] [Figure 6] This document presents an exemplary active learning system that can be used to implement one or more aspects of various embodiments.

[0010] [Figure 7] An exemplary decision tree is shown illustrating a method for determining manufacturability scores using peptide-derived factors in various embodiments.

[0011] [Figure 8] This document illustrates an exemplary method that can be used according to various embodiments.

[0012] [Figure 9] This shows the components of an exemplary computing device that can be used according to various embodiments.

[0013] [Figure 10] Examples of environments for implementing various embodiments are shown.

[0014] [Figure 11] This shows components of another exemplary environment that can implement various embodiments.

[0015] [Figure 12] A and B are exemplary histograms showing the expected number of peptides produced based on a manufacturing simulation process in various embodiments.

[0016] [Figure 13] A and B are exemplary histograms showing a comparison between manufacturing simulation methods optimized for different criteria according to various embodiments. **Mode for Carrying Out the Invention**

[0017] In the following description, various embodiments will be described. For the purpose of explanation, specific configurations and details have been made clear in order to provide a complete understanding of the embodiments. However, it will be apparent to those skilled in the art that the embodiments can be implemented without specific details. Furthermore, well-known features may be omitted or simplified in order not to obscure the described embodiments.

[0018] A manufacturer may provide a list of heuristics or criteria that can be used to determine whether a peptide is manufacturable. Manufacturability, as used herein in the context of peptide synthesis, may refer to the process of predicting whether a peptide can be practically and reliably manufactured for research, therapeutic, or commercial purposes. In one embodiment, manufacturability may be determined by a process of evaluating and / or predicting one or more characteristics related to the peptide, which may include one or more of synthesizable, solubility, quality prediction, and other factors that may affect the production of the peptide. For example, a manufacturer may indicate that a peptide should not have too many hydrophobic amino acids. However, if there are amino acids that increase hydrophobicity but also increase stiffness (due to the fact that stiff peptides may be more likely to be soluble in certain solvents), a decision can be made regarding whether the trade-off of using those amino acids for manufacturing is worthwhile. For example, in aqueous solutions (water-based), increased hydrophobicity usually decreases solubility, but increased stiffness may offset this effect to some extent by stabilizing the peptide structure. In other solvents, such as nitrogen-based solvents (e.g., nitromethane) or organic solvents (e.g., ethanol), the effects of hydrophobicity and stiffness may differ. Stiffness can either enhance or decrease solubility depending on the specific chemical interactions between the peptide and the solvent. Therefore, the selection of amino acids for peptide synthesis may require consideration of the intended solvent environment, which may include an assessment of both hydrophobicity and stiffness regarding solubility and overall manufacturability in various solvent systems. The main factor contributing to stiffness is the presence of the amino acid proline, which has a pentatomic ring, making it inherently much less flexible than other amino acids. Since stiff peptides cannot easily bend to form bonds with other peptides, it is hypothesized that proline (and other amino acids that make peptides stiffer) keeps the peptide in solution (i.e., soluble). If peptides aggregate, they tend to detach from the solution, lacking manufacturability and therefore unusable in the final product.Furthermore, since the solution containing a given peptide can vary depending on the use case, its solubility can also vary depending on the use case. Moreover, considering the countless potential peptide sequences, each possessing its own unique characteristics that must be considered for analysis, accurately determining the trade-offs without using computer-based algorithms is difficult, if not impossible. For example, there may be thousands of sequences for synthesis, each with hundreds of features for analysis, which would be data overload for humans to predict at the level of accuracy required for manufacturing. Furthermore, using computer-based algorithms allows manufacturability to be evaluated in a completely objective manner with little room for error. Additionally, manufacturers may be required to create mixtures of peptides in a particular solution. In one embodiment, individual peptides may be independently soluble but may aggregate and become unusable when combined in solution. Computer-based algorithms (e.g., machine learning algorithms) can be used to select co-soluble peptides.

[0019] Figure 1 shows an exemplary method 100 that can be used according to various embodiments. According to various embodiments, a request for information related to the manufacturability of a peptide can be received. A determination of whether a peptide is predicted to be synthesizable can be made, for example, by using a machine learning model. The machine learning model can be trained on data including one or more of the following: manufacturer specifications, descriptions related to the peptide, and peptide characteristics. A second determination can be made regarding whether a peptide is predicted to be soluble, using the same or different machine learning model trained on peptide solubility data. If a peptide is predicted to be soluble and synthesizable, a manufacturability score for the peptide can be determined. The manufacturability score may correspond to or indicate the likelihood of the peptide being manufacturable. In some exemplary embodiments, a determination can also be made regarding whether a peptide is likely to pass quality checks that have conventionally been determined using mass spectrometry techniques. Quality checks ensure that the peptide actually produced is the originally requested peptide. Individual predictions of whether a peptide is synthesizable, soluble, and will pass quality checks can be made using the same or different machine learning models. Figure 1 shows an exemplary sequence of different stages of prediction and evaluation, but it should be understood that the sequence in which these evaluations are performed may vary or occur simultaneously. The order of operations may vary depending on the specific manufacturing context or data availability, regardless of whether synthesizability, solubility, quality, or other characteristics are being predicted.

[0020] The example shown in FIG. 1 is such that the generation requirements of peptide 110 or peptide representation can be input into a model such as a hierarchical model to evaluate manufacturability by a manufacturer. The synthesizability 120 of the peptide can be evaluated. If the peptide is predicted to be synthesizable, then it can be determined whether the peptide passes the quality control check 130. If it is determined that the peptide is not predicted to be synthesizable, the analysis regarding manufacturability fails and a false response 160 is associated. As used herein, synthesizability can indicate the feasibility and efficiency of being chemically synthesized based on the amino acid sequence and structural characteristics of the peptide. In one embodiment, synthesizability can depend on the practicality of chemically generating the peptide based on its amino acid sequence and structural characteristics. For example, factors affecting synthesizability may include the specific amino acids involved, which can affect the stability and reactivity of the peptide during synthesis. In one embodiment, synthesizability can also depend on the thermal stability of the peptide, its reaction to pH variations, or its interaction with other molecular components of the formulation. In another example, sequences that are prone to forming complex secondary structures, or sequences containing chemically sensitive residues such as cysteine or asparagine, can complicate the synthesis and purification processes. Further, longer sequences can face a high incidence of synthesis errors and challenges during purification, so the length of the peptide chain can also be a factor.

[0021] In the quality control phase, the process may include using a predictive model to estimate whether the synthesized peptide meets specified quality criteria. This model-based prediction can serve as an initial pre-production check to assess the likelihood of the peptide passing quality control. In some embodiments, after the peptide is produced, its quality may be verified using mass spectrometry techniques. Such empirical verification can not only confirm the model's predictions but also generate essential data that can be used to further refine the predictive model. If the model predicts that the peptide will fail the quality check, and this is confirmed by mass spectrometry, the assessment of the peptide's manufacturability is considered a failure, resulting in a false response 160 in the model's output. In one embodiment, the determination may be indicated by data generated through mass spectrometry techniques. Mass spectrometry can be used to analyze the composition and structure of the synthesized peptide, providing high-quality data to be used to train the predictive model. By incorporating mass spectrometry data, the model learns to accurately predict the quality of the peptide, ensuring that the produced peptide matches the peptide initially specified in the design.

[0022] If the synthesized peptide passes quality control, it can be determined whether the synthesized peptide is soluble in solution.140 In one embodiment, solubility may be influenced by the interaction between the chemical characteristics of the peptide and the solvent used. The choice of solvent can significantly affect solubility, as peptides composed of hydrophilic amino acids such as lysine and arginine are generally soluble in polar solvents such as water, while peptides with hydrophobic amino acids such as leucine and valine may dissolve better in nonpolar solvents. For example, to promote dissolution, the polarity of the solvent should match the polarity of the peptide. In one embodiment, solubility may also be influenced by the length of the peptide, with shorter chains tending to be more soluble, and by the ionic strength and pH of the solution, which can change solubility by affecting the charge of the peptide. If the synthesized peptide is predicted to be soluble, the required peptide can be manufactured using the true value.150 If the synthesized peptide is not predicted to be soluble, the manufacturability analysis fails, and a false response is associated.160 By adding false responses to the model's training dataset, the dataset can be augmented, improving the prediction of peptide manufacturability for future demands using active learning techniques.

[0023] In one embodiment, such a hierarchical model is particularly useful when the dataset may contain only a small number of samples. If a sample is determined to be insoluble, unsynthetic, or fails quality checks, that data is typically excluded and not used in any meaningful way. By using a hierarchical model, all of the data can be used to inform a better decision by the model as a whole. Each individual decision on synthesizability 120, quality 130, and solubility 140 can each be its own intermediate model that informs a larger machine learning model about the manufacturability of the requested peptide 110. For example, positive data corresponding to decisions that yield favorable results can also be added to the training dataset. Since solubility and synthesizability data are often unknown for a given peptide, the active learning process of the hierarchical model can improve the prediction of manufacturability over time.

[0024] Binary prediction tasks that predict manufacturability can be broken down into more granular decisions in this way, and relevant labels can be inferred based on the hierarchy of tasks. For example, if a peptide is successfully manufactured, all subtasks (e.g., synthesizability, solubility, and quality checks) can be associated with positive labels. In some cases, all labels may be unknown. For example, if a peptide fails at an early stage, the manufacturer may not test it at a later stage.

[0025] The hierarchical models described herein allow different subtasks to be trained separately (e.g., using different models for each individually trained subtask) or jointly (e.g., using a single multitask-trained model for all subtasks). Training of three subtasks and one parent task can be done using four binary classifiers, with one binary classifier per task. The subtasks may correspond to predictions of synthesizability, quality checks, and solubility of a given peptide, while the parent task may correspond to predictions of peptide manufacturability. Alternatively, a multi-class classifier can be used to cover all hierarchical options, where classes indicate whether the peptide failed during synthesis prediction, passed synthesis prediction but failed quality checks, passed synthesis prediction and quality checks but failed solubility prediction, and passed all stages. Depending on the specific context and / or different factors, different stages can be tested in any order.

[0026] Figure 2 shows an exemplary representation of a manufacturability analysis system that can be used according to various embodiments. In this example, a set of peptide samples 210a, 210b, 210c, and 210n may be required for analysis and manufacturing. Although this example shows only four peptide samples, any number of "N" samples, such as thousands of peptide samples, may be required. Each peptide sample may have hundreds of features (e.g., "220a-n, 222a-n, 224a-n, 226a-n" in Figure 2) generated considering the manufacturability analysis module 230. Features include, but are not limited to, amino acid sequence determination information, length, thermal stability, molecular weight, charge, polarity, hydrophobicity, hydrogen bonding, number of specific amino acids, isoelectric properties, topological descriptors, secondary structure, stiffness, conformation, post-translational modifications, other heuristics affecting manufacturability, and meta-properties derived from dimensionality reduction strategies of the principal components (and related). These features can be used to describe the overall peptide composition, the N and C-terminal amino acids of the peptide, and the minimum / maximum values ​​of the fixed sliding window applied to the peptide. Overall, there are hundreds of features and descriptors that can be analyzed for each peptide.

[0027] The manufacturability analysis module 230 can analyze peptides and their associated characteristics to assess their manufacturability. The manufacturability analysis module 230 can evaluate characteristics such as synthesizability, quality control, and solubility, as described herein with respect to Figure 1. When each peptide involves complex data analysis involving numerous input peptides and characteristics, the manufacturability analysis module 230 can propose decisions regarding the manufacturability of the peptide in an efficient and objective manner, improving accuracy and enabling intelligent decision-making for peptide production. For example, when amino acids are added to a peptide that increases both hydrophobicity and stiffness, the underlying model of the manufacturability analysis module 230 can accurately determine how much the hydrophobicity and stiffness increase, providing a better judgment on whether the peptide is worth manufacturing. Laboratory technicians may not be able to determine such information with the same level of precision and accuracy as the model.

[0028] Figures 3A and 3B show exemplary histograms 300a and 300b for predicting manufacturability according to various embodiments. Figures 3A and 3B show peptide manufacturability predictions visualized through two different versions of the prediction model.

[0029] In Figure 3A, the horizontal axis 320a represents the predicted manufacturability score in the range of 0.0 to 1.0, with scores closer to 1.0 indicating a higher probability of successful manufacturing. The vertical axis 310a shows the peptide count, with the number of peptides predicted to fall within each score in parentheses. In the histogram, the open bars represent the total number of peptides subjected to the manufacturing process based on these predictions, while the bars with shaded areas inside represent the number of peptides that were actually successfully manufactured. Figure 3A presents a visual representation of the success of predictions relative to actual results. The arrangement or overlap between the open bars and the shaded areas inside these bars indicates the effectiveness of the model. If the shaded areas occupy a large portion of the open bars, most of the peptides in that score bin were successfully manufactured. For a good manufacturability predictor, the proportion of open bars covered by shaded areas increases as the manufacturability score increases. In one embodiment, such a model may utilize a "leave-one-out" (LOO) cross-validation process to ensure accuracy, particularly in data-limited scenarios. The LOO process can most accurately demonstrate the quality of the model by using each data point in the dataset as a test set one at a time, while the rest of the data serves as the training set. This method is beneficial in data-scarce scenarios because it maximizes the use of data available for training and testing, resulting in a complete assessment of the model's predictive power.

[0030] Figure 3B shows another histogram corresponding to an updated model with more descriptors for improved accuracy. The horizontal axis 320b may represent manufacturability scores from 0.0 to 1.0, and the vertical axis 310b may represent the number of peptides per predicted score. Hollow-outlined bars represent the total amount of peptides processed, and the shaded areas represent peptides that were successfully manufactured. In one embodiment, exemplary histogram 300b considers additional manufacturability data and uses a "Least Absolute Shrinkage and Selection Operator (LASSO)" regression model and strict LOO cross-validation. In one embodiment, LASSO regression is a type of linear regression that performs both variable selection and regularization to improve the predictive accuracy and interpretability of the statistical model it produces. In one embodiment, the underlying model shown in Figure 3B may contain more descriptors than the model corresponding to Figure 3A. When dealing with datasets containing a large number of predictors (variables), LASSO regression can be used to penalize the absolute size of the regression coefficients. The underlying model may consider features such as the number of certain types of amino acids at the c-terminus or n-terminus of a peptide.

[0031] Figure 4 shows exemplary receiver operating characteristic (ROC) curves 412 for predicting solubility in various embodiments. In one embodiment, the ROC curve 412 is used to evaluate the performance of a predictive model in terms of its ability to accurately classify results. The curve plots the true positive rate (TPR) on the vertical axis 410 against the false positive rate (FPR) on the horizontal axis 420, which shows the trade-off between benefits (true positives) and costs (false positives). TPR, also known as sensitivity, indicates the model's ability to accurately identify actual positives. In contrast, FPR measures the proportion of negatives that are incorrectly identified as positives. The ROC curve 412 in this example uses the same data as the histograms in Figures 3A and 3B. The true positive rate 410 and false positive rate 420 can be analyzed against the manufacturability score 430 as they move across various manufacturability scores 430 at a given point in time. The effectiveness of the model is generally measured by the area under the ROC curve (AUC). For example, a perfect model would score an AUC of 1, indicating that it completely distinguishes between peptides that can and cannot be synthesized. From a practical standpoint, analyzing this curve allows for the selection of an appropriate threshold score that maximizes the accuracy of the model, ensuring that the manufacturing process concentrates resources on peptides that are most likely to be successfully synthesized, while avoiding consumption on peptides that will not be successfully synthesized. In Figure 4, the manufacturability score of the ROC curve 412 is shaded with an Area Under the Curve (AUC) value of approximately 0.80. The true positive rate 410 represents a measure of the proportion of actual positive cases that were successfully manufactured. The false positive rate 420 is a measure of the proportion of actual negative cases that were not manufactured.

[0032] Figure 5 presents a calibration curve that converts the model score to the predicted probability of peptide manufacturability. This curve allows for the interpretation of the model output, which may present a score that is not directly expressed as a percentage. The horizontal axis represents the model score 520, displaying the score generated by the model, which may be in the range of 0 to 1 in this example. In one embodiment, the model score may be a numerical value in a different range, e.g., 0 to 12 in other model variations. For such models, a conversion from the model score to the predicted probability is necessary (e.g., converting a model score of 11.4 to a probability). The vertical axis represents the predicted probability of manufacturability 510, indicating the predictability that the peptide can be successfully manufactured based on the model output.

[0033] Figure 6 shows an exemplary active learning system 600 that can be used to implement one or more aspects of various embodiments. Active learning, as applied to various embodiments described herein, can be referred to as the underlying process of the model for selecting the peptide for manufacturing that best improves the model. For example, peptide X may have an 80% manufacturability score, while peptide Y may have a 60% manufacturability score. Regardless of these scores, peptide X may have a “specificity” score of 1, meaning that the model is certain that peptide X has an 80% manufacturability score. In contrast, peptide Y may have a “specificity” score of 9, indicating a high degree of uncertainty that peptide Y has a 60% manufacturability score. In the case of a peptide with high uncertainty, such as peptide Y in the above example, peptide Y can be manufactured, and the manufacturability result can be provided to a model, such as a machine learning model. Furthermore, according to one or more embodiments described herein, a particular peptide may be requested for manufacturing in order to specifically improve the training dataset.

[0034] In this example, the training dataset 610 may be associated with the model 620. The model 620 may be a machine learning model trained to predict, or otherwise assess, the manufacturability of a peptide. The training data 610 may include, but is not limited to, peptides and peptide representations, peptide features or heuristics / criteria, manufacturer specifications and descriptors, and solubility. The training data can be correlated using a single value that represents a particular aspect of the training data. For example, heuristics or criteria can be correlated using a value or set of values ​​that represent the heuristic or criteria. The model 620 may include, but is not limited to, a synthesizability analysis module 630, a quality control analysis module 640, and a solubility analysis module 650. The synthesizability analysis module 630 can be used, according to one or more embodiments, to assess whether a peptide in the input data is predicted to be synthesizable. If a peptide is predicted to be synthesizable, the model can determine whether the peptide is likely to pass quality control checks using the quality control analysis module 640. For quality control, the model can determine whether the synthesizable peptide is likely to conform to a required peptide. In one embodiment, the model may use a machine learning model to predict whether a peptide is likely to conform to the required peptide. After peptide synthesis, in some exemplary embodiments, the quality of the synthesized peptide can be verified using mass spectrometry techniques, and the mass spectrometry is used as training data when training a model to predict synthesizability. If a synthesizable peptide is predicted to pass quality control, it can be determined whether the synthesizable peptide is predicted to be soluble in solution using the solubility analysis module 650. If a synthesizable peptide is predicted to be soluble, peptide 660 may be intended to be manufactured. Individual predictions about whether a peptide is synthesizable, soluble, and will pass quality checks can be made using the same or different machine learning models. The order of individual predictions may differ based on the specific context.Furthermore, positive data corresponding to decisions that yield favorable outcomes can also be associated with the machine learning model(s) and added to the training dataset.

[0035] In some cases, a peptide may be predicted not to be synthesizable, fail predicted quality checks, be predicted to be insoluble, or otherwise be judged as unmanufacturable. These results 670 can be added to the training dataset to expand the dataset and improve the prediction of peptide synthesizability for future demands using active learning techniques. Such hierarchical models are particularly useful when the dataset may have only a small number of samples. If a sample is judged to be insoluble or fails quality checks, that data is usually excluded and not used in any meaningful way. By using a hierarchical model, all of the data can be used to inform a better decision by the model as a whole. The synthesizability analysis module 630, the quality control analysis module 640, and the solubility analysis module 650 can all be their own intermediate models that inform a larger machine learning model about the manufacturability of the requested peptide. Since solubility data is often unknown for a given peptide, the active learning process of the hierarchical system can improve the prediction of manufacturability over time. Figure 7 shows an exemplary decision tree 700 that can be used according to various embodiments. Decision trees, such as the decision tree 700 in Figure 7, can provide additional insights into why a peptide is or is not synthesizable. Various amino acid (AA) properties or characteristics ("AA property 1," "AA property 2," "AA property 3," "AA property 4") can be analyzed against various thresholds ("low" and "high"). According to exemplary embodiments, AA properties may include, but are not limited to, hydrophobicity, charge, side-chain properties, acid-base properties, stereochemistry, hydrogen bonding, and chemical reactivity. Based on the data given in the decision tree 700, it can be determined how likely it is that a soluble peptide can be successfully synthesized. The decision tree 700 can provide information at various nodes ("node 1," "node 2," "node 3," "node 4," "node 5") regarding the likelihood of successful peptide synthesis. For example, if AA property 4 is low but AA property 3 is high, the likelihood of successful peptide synthesis is higher than if AA property 3 is also low.If AA characteristic 4 is high, AA characteristic 2 is low, and AA characteristic 1 is also low, the probability of successful peptide synthesis is very high. Decision trees can provide insight into which specific characteristics can lead to successful peptide synthesis, which is something that engineers cannot easily determine by other methods, given a vast number of features and amino acid sequences to analyze.

[0036] Figure 8 shows an exemplary method 800 that can be used according to various embodiments. It should be understood that, for any process described herein, unless otherwise specified, additional, fewer, or alternative steps may exist within the scope of various embodiments, performed in a similar or alternative order, or in parallel. A request for information relating to the manufacturability of a peptide can be received 810. A determination of whether a peptide is synthesizable can be made, for example, by using a machine learning model 820. The machine learning model can be trained with data including manufacturer specifications, descriptions related to the peptide, and characteristics of the peptide. A second determination can be made regarding whether a peptide is soluble 830, using the same or a different machine learning model trained with peptide solubility data. If the peptide is soluble and synthesizable, a manufacturability score for the peptide can be determined 840. The manufacturability score may correspond to or indicate the likelihood of successfully manufacturing the peptide. In some exemplary embodiments, a determination can also be made regarding whether a peptide is predicted to pass quality checks that can be determined using mass spectrometry techniques. For example, the determination may be informed by data generated through mass spectrometry techniques. This process involves using mass spectrometry to analyze the composition and structure of the synthesized peptide, providing high-quality data used to train a predictive model. By incorporating the mass spectrometry data, the model learns to accurately predict the quality of the peptide, ensuring that the produced peptide matches the peptide initially specified in the design. Quality checks help ensure that the peptide actually produced is the peptide originally requested.

[0037] According to some exemplary embodiments, instead of explicitly extracting the biochemical features of peptides, deep learning can be used to predict the biochemical features of peptides by using embeddings of pre-trained peptide / protein "language models," protein structure prediction models, or combinations thereof. The deep learning model may have a hidden layer for each peptide embedding, and transfer learning can be used to perform the novel task of acquiring the embedded model, transferring it, and using the embedded model to make predictions.

[0038] According to one or more embodiments, a pre-trained protein language model, such as Evolutionary Scale Modeling (ESM), may be used to extract information peptide embeddings used to train a manufacturability predictive classifier. Alternatively, after fine-tuning on a large peptide dataset, embeddings based on a protein model or a structure prediction model may be used. Embeddings derived from a pre-trained model can be complemented by known biochemical or physiological properties derived directly from the amino acid sequence and used as input peptide representations for the manufacturability classifier.

[0039] Computing resources such as servers that can update software and / or firmware in such matters generally include at least a set of standard components configured for general-purpose operation, although various proprietary components and configurations may also be used within the scope of various embodiments. Figure 9 shows components of an exemplary computing device 900 available according to various embodiments. As is known about computing devices, a computer has one or more processors 902, such as a central processing unit (CPU) and a graphics processing unit (GPU), which are electronically and / or communicably coupled with various components using various buses, traces, and other such mechanisms. A processor 902 may include memory registers 906 and cache memory 904 for holding instructions, data, etc. In this example, a chipset 914, which may include a northbridge and a southbridge in some embodiments, can work with various system buses to connect the processor 902 to components such as system memory 916 in the form of physical RAM or ROM, which may include operating system code, as well as various other instructions and data used for the operation of the computing device. The computing device may also include, or communicate with, one or more storage devices 920, such as hard drives, flash drives, and optical storage, for persisting data and instructions, as well as data and instructions stored in the processor and memory. The processor 902 may also communicate with various other components via the chipset 914 and interface buses (or graphics buses, etc.), which may include communication devices 924, such as cellular modems or network cards; media components 926, such as graphics cards and audio components; and peripheral interfaces 930 for connecting peripherals such as printers and keyboards.At least one cooling fan 932 or other such temperature control or reduction component may also be included, which may be driven by the processor or triggered by various other sensors or components on or off the device. Various other or alternative components and configurations can be utilized, as is well known in the art with respect to computing devices.

[0040] In some embodiments, at least one processor 902 can retrieve data from physical memory 916, such as a dynamic random access memory (DRAM) module, via a coherency fabric. It should be understood that various architectures can be utilized in such computing devices, and within the scope of various embodiments, various choices, numbers, and arguments of buses and bridges may be involved. Memory data can be managed and accessed via the coherency fabric by a memory controller, such as a DDR controller. In at least some embodiments, the data may be temporarily stored in a processor cache 904. Furthermore, the computing device 900 can support multiple I / O devices using a set of I / O controllers connected via an I / O bus. Among other options, there may be I / O controllers supporting various types of I / O devices, such as Universal Serial Bus (USB) devices, data storage (e.g., flash or disk storage), network cards, Peripheral Component Interconnection Express (PCIe, etc.) cards or interfaces 930, communication devices 924, graphics or audio cards 926, and direct memory access (DMA) cards. In some embodiments, components such as the processor, controller, and cache can be configured on a single card, board, or chip (i.e., a system-on-a-chip embodiment), while in other embodiments, at least some of the components are located in different locations, and so on.

[0041] The operating system (OS) running on the processor 902 can help manage various devices that may be used to provide input to be processed. This may include, for example, enabling interaction with various I / O devices using associated device drivers, which may relate to data storage, device communication, user interfaces, etc. Various I / O devices are typically connected via various device ports and communicate with the processor and other device components via one or more buses. Among other options, there may be specific types of buses that provide communication according to specific protocols, such as Peripheral Interconnect (PCI) or Small Computer System Interface (SCSI) communication. Communication may occur using registers associated with each port, including data-in and data-out registers. Communication may also be performed using memory-mapped I / O, in which a portion of the processor's address space is mapped to a specific device, and data is written directly to and from that portion of the address space.

[0042] Such devices may be used, for example, as servers in a server farm or data warehouse. Server computers often need to perform tasks outside the environment of the CPU and main memory (i.e., RAM). For example, a server may need to communicate with an external entity (e.g., another server) or process data using an external processor (e.g., a general-purpose graphical processing unit (GPGPU)). In such cases, the CPU may interface with one or more I / O devices. In some cases, these I / O devices may be dedicated hardware designed to perform a specific role. For example, an Ethernet® network interface controller (NIC) may be implemented as an application-specific integrated circuit (ASIC) containing digital logic capable of transmitting and receiving packets.

[0043] In exemplary embodiments, the host computing device is associated with various hardware and software components and their respective configurations that facilitate the execution of I / O requests. One such component is an I / O adapter that inputs and / or outputs data along a communication channel. In one embodiment, the I / O adapter device can communicate as a standard bridge component to facilitate access between various physical and emulated components and the communication channel. In another embodiment, the I / O adapter device may include an embedded microprocessor, which can then execute computer executable instructions related to the management of an embodiment of a management function or one or more such management functions, or other computer executable instructions related to an embodiment of the I / O adapter device. In some embodiments, the I / O adapter device may be implemented using multiple separate hardware elements, such as multiple cards or other devices. The management controller may be configured in such a way that it is electrically isolated from any other components of the host device other than the I / O adapter device. In some embodiments, the I / O adapter device is external to the host device. In some embodiments, the I / O adapter device is internally integrated into the host device. Furthermore, in communication with the I / O adapter device, an external communication port component may exist to establish a communication channel between the host device and one or more network-based services, or other network-connected or directly connected computing devices. Exemplarily, the external communication port component may correspond to a network switch, sometimes known as a top-of-rack ("TOR") switch. The I / O adapter device can utilize the external communication port component to maintain a communication channel between the host device and one or more services, such as health check services and financial services.

[0044] I / O adapter devices can also communicate with Basic Input / Output System (BIOS) components. BIOS components may contain non-temporary executable code, often called firmware, which may be executed by one or more processors and is used to cause components of a host device to initialize and identify system devices such as video display cards, keyboards and mice, hard disk drives, optical disc drives, and other hardware. BIOS components may also contain or place boot loader software used to boot the host device. For example, in one embodiment, a BIOS component may contain executable code, which, when executed by a processor, attempts to cause the host device to locate pre-boot execution environment (PXE) boot software. Furthermore, BIOS components may include, or utilize, hardware latches electrically controlled by the I / O adapter device. Hardware latches can restrict access to one or more aspects of the BIOS component, such as controlling changes to or configuration of the executable code maintained by the BIOS component. BIOS components may connect to (or communicate with) several additional computing device resource components, such as processors and memory. In one embodiment, such computing device resource components may be physical computing device resources that communicate with other components via communication channels. The communication channel can support one or more communication buses, such as a shared bus (e.g., processor bus, memory bus) or a point-to-point bus such as PCI or PCI Express Bus, over which components of a bare-metal host device communicate. Other types of communication channels, communication media, communication buses, or communication protocols (e.g., Ethernet® communication protocol) may also be used.Furthermore, in other embodiments, one or more of the computing device resource components may be virtualized hardware components emulated by the host device. In such embodiments, the I / O adapter device can perform management processes in which the host device consists of physical or emulated hardware components, based on various criteria. The computing device resource components may communicate with the I / O adapter device via a communication channel. Furthermore, the communication channel may connect the PCI Express device to the CPU via a northbridge or host bridge, among other options.

[0045] When communicating with an I / O adapter device via a communication channel, one or more controller components may exist to manage hard drives or other forms of memory. An example of a controller component might be a SATA hard drive controller. Similar to a BIOS component, a controller component may include, or benefit from, hardware latches that are electrically controlled by the I / O adapter device. Hardware latches can restrict access to one or more aspects of the controller component. Exemplarily, hardware latches may be controlled together or independently. For example, an I / O adapter device may selectively close the hardware latches of one or more components based on a trust level associated with a particular user. In another example, an I / O adapter device may selectively close the hardware latches for one or more components based on a trust level associated with the author or seller of executable code executed by the I / O adapter device. In a further example, an I / O adapter device may selectively close the hardware latches of one or more components based on a trust level associated with the component itself. A host device may also include additional components that communicate with one or more exemplary components associated with the host device. Such components may include devices such as one or more controllers combined with one or more peripheral devices such as hard disks or other storage devices. Furthermore, additional components of the host device may include another set of peripheral devices, such as graphics processing units ("GPUs"). Peripheral devices may also be associated with hardware latches to restrict access to one or more aspects of the components. As described above, in one embodiment, the hardware latches may be controlled together or independently.

[0046] As described, different approaches can be implemented in various embodiments according to the embodiments described. For example, Figure 10 shows an example environment 1000 for implementing embodiments according to various embodiments. As can be understood, a web-based environment is used for explanatory purposes, but different environments may be used as appropriate to implement various embodiments. The system includes an electronic client device 1002, which may include a suitable device that can be operated to send and receive requests, messages, or information over a suitable network 1004 and to transmit and return information to the user of the device. Examples of such client devices include personal computers, mobile phones, portable messaging devices, laptops, set-top boxes, personal digital assistants, e-book readers, etc. The network may include any suitable network, including intranets, the internet, cellular networks, local area networks, or any other such networks, or combinations thereof. The components used in such a system may depend, at least to some extent, on the type of network and / or the environment chosen. Protocols and components for communication over such networks are well known and will not be discussed in detail herein. Communication over the network may be possible via wired or wireless connections, and combinations thereof. In this example, the network includes the Internet because the environment includes a web server 1006 for receiving requests and providing content in response; however, for other networks, alternative devices that serve a similar purpose may be used, as may be obvious to those skilled in the art.

[0047] The exemplary environment includes at least one application server 1008 and a data store 1010. It should be understood that there may be multiple application servers, layers, or other elements, processes, or components that are linked or otherwise configured and can interact in such a way as to perform tasks such as retrieving data from a suitable data store. As used herein, the term “data store” refers to any device or combination of devices capable of storing, accessing, and retrieving data, which may include any combination and any number of data servers, databases, data storage devices, and data storage media in any standard, distributed, or clustered environment. The application server 1008 interacts with the data store 1010 as needed to execute one or more aspects of an application relating to a client device and may include any suitable hardware and software to handle the majority of the data access and business logic relating to the application. The application server may work with the data store to provide access control services and generate content such as text, graphics, audio, and / or video, which will be delivered to the user, and which in this example may be served to the user by the web server 1006 in the form of HTML, XML, or another suitable structured language. The processing of all requests and responses, as well as the delivery of content between the client device 1002 and the application server 1008, can be handled by the web server 1006. Since the structural code described herein can be executed on any suitable device or hosted on a machine as described elsewhere herein, it should be understood that the web and application servers are not required and are merely illustrative components.

[0048] The data store 1010 may include multiple separate tables, databases or other data storage mechanisms, and media for storing data relating to a particular aspect. For example, the data store described may include a mechanism for storing peptide 1012 or peptide representation and analysis data 1016, which may be used to provide content to the production side. The data store may also include a mechanism for storing feature data 1014 for the peptide. There may be numerous other aspects that need to be stored in the data store, such as manufacturing data or other training data, which can be stored in any of the mechanisms listed above or in additional mechanisms within the data store 1010, as needed. The data store 1010 can receive commands from the application server 1008 through logic related to the data store and operate to retrieve, update, or otherwise process data in response to the commands. In one example, a user may submit a request for peptide synthesis. In this case, the data store may be used to access peptide information, feature data, and analysis data to obtain information on whether the peptide is synthesizable. This information may then be returned to the user, for example, in a results list on a web page that the user can view via a browser on the user device 1002.

[0049] Each server typically includes an operating system that provides executable program instructions for the general management and operation of the server, and typically includes a computer-readable medium that, when executed by the server's processor, stores instructions that cause the server to perform its intended functions. Preferred implementations of the server's operating system and general functionality are known or commercially available and, in particular, readily implementable by those skilled in the art in light of the disclosures herein.

[0050] The environment in one embodiment is a distributed computing environment that utilizes multiple computer systems and components interconnected via communication links, using one or more computer networks or direct connections. However, it will be understood by those skilled in the art that such a system may function equally well in a system with fewer or more components than illustrated in Figure 10. Accordingly, the depiction of system 1000 in Figure 10 should be considered illustrative in nature and not limiting the scope of this disclosure.

[0051] Figure 11 illustrates an example environment 1100 that can implement various embodiments. In this example, a user can utilize a client device 1102 to submit requests to a multitenant resource provider environment 1106 over at least one network 1104. The client device may include any suitable electronic device capable of transmitting and receiving requests, messages, or other such information over a suitable network and relaying the information back to the user of the device. Examples of such client devices include personal computers, tablet computers, smartphones, and notebook computers. The at least one network 1104 may include any suitable network, including an intranet, the Internet, a cellular network, a local area network (LAN), or any other such network, or a combination thereof, and communication over the network may be enabled via wired and / or wireless connections. The resource provider environment 1106 may include any suitable components for receiving requests, returning information in response to those requests, or performing actions. For example, the provider environment may include web servers and / or application servers for receiving and processing requests and then, in response to those requests, returning data, web pages, videos, audio, or other such content or information.

[0052] In various embodiments, the provider environment may include various types of resources that can be used by multiple users for various different purposes. As used herein, computing and other electronic resources used in a network environment may be referred to as “network resources.” These may include, for example, servers, databases, load balancers, routers, etc., that can perform tasks such as receiving, transmitting, and / or processing data and / or executable instructions. In at least some embodiments, all or some of a given resource, or a set of resources, may be assigned to a particular user for at least a defined period of time, or to a particular task. The sharing of those multitenant resources from the provider environment is often referred to as resource sharing, web services, or “cloud computing,” in other such terms, and depending on the particular environment and / or implementation. In this example, the provider environment includes multiple resources 1114 of one or more types. These types may include, for example, application servers capable of operating to process instructions provided by a user, or database servers capable of operating to process data stored in one or more data stores 1116 in response to user requests. As is known with respect to such processes, users can also reserve at least a portion of the data storage in a given datastore. Methods for enabling users to reserve various resources and resource instances are well known in the art, and as a result, a detailed description of the entire process and all possible components is not discussed in detail herein.

[0053] In at least some embodiments, a user who wishes to utilize a portion of the resources 1114 can submit a request to the interface layer 1108 of the provider environment 1106. The interface layer may include an application programming interface (API) or other exposed interface that enables the user to submit requests to the provider environment. The interface layer 1108 in this example may also include other components such as at least one web server, a routing component, and a load balancer. When a request to provision resources is received by the interface layer 1108, information about the request may be directed to the resource manager 1110, or to any other such system, service, or component configured to manage user accounts and information, resource provisioning and usage, and other such aspects. The resource manager 1110 receiving the request may perform tasks such as authenticating the identity of the user submitting the request and determining whether the user has an existing account with the resource provider, and the account data may be stored in at least one data store 1112 within the provider environment. The user may provide the provider with any of various types of authentication credentials to authenticate the user's identity. These authentication credentials may include, for example, username and password pairs, biometric data, digital signatures, or other such information. The provider can verify the legitimacy of this information against the information stored for the user. If the user has an account with appropriate permissions, status, etc., the resource manager can determine if there are suitable available resources that meet the user's request, and if so, can provision the resources or grant the user access to the corresponding portion of those resources for use in the amount specified by the request.This quantity may include, for example, the ability to process a single request or perform a single task, a specified period of time, or a recurrence / reconfiguration period, among other such values. If a user does not have a valid account with the provider, and the user account does not allow access to the type of resource specified in the request, or for other such reasons, a communication may be sent to the user to allow the user to create or modify an account or modify the resource specified in the request, among other such options.

[0054] Once a user is authenticated, their account is verified, and resources are allocated, the user can utilize the allocated resources(s) for a specified capacity, amount of data transfer, duration, or other such values. In at least some embodiments, the user may provide a session token or other such authentication information with subsequent requests to enable those requests to be processed within the user session. The user may receive a resource identifier, a specific address, or other such information, which may enable the client device 1102 to communicate with the allocated resources without needing to communicate with the resource manager 1110 until at least the relevant aspect of the user account changes, until the user is no longer permitted to access the resources, or until another such aspect changes, etc.

[0055] In this example, the resource manager 1110 (or another such system or service) can also function as a virtual layer of hardware and software components that handle control functions in addition to management actions, such as provisioning, scaling, and replication. The resource manager may have access to dedicated APIs in the interface layer 1108, each of which may be provided to receive requests for at least one specific action to be performed on the data environment, such as provisioning, scaling, cloning, or hibernating instances. Upon receiving a request to one of the APIs, the web service portion of the interface layer may syntactically analyze the request to determine the steps or actions required to act on or process the call, or otherwise analyze it. For example, a web service call may be received that includes a request to create a data repository.

[0056] In at least one embodiment, the interface layer 1108 includes an extensible set of user-facing servers that can provide various APIs and return appropriate responses based on API specifications. The interface layer may also include, in one embodiment, at least one API service layer consisting of stateless, replicated servers that handle external user APIs. The interface layer may be involved with web service frontend functions such as authenticating users based on credentials, granting users authorizations, filtering user requests to API servers, verifying the validity of user input, and marshalling or unmarshalling requests and responses. The API layer may also be involved with reading and writing database configuration data to and from the administration data store in response to API calls. In many embodiments, the web service layer and / or the API service layer are the only components visible from the outside, or visible to users of the control service and accessible by the customer. The servers in the web service layer may be stateless and horizontally scalable, as is known in the art. The API server and persistent data store may be distributed, for example, across multiple data centers within a given area, thereby providing resilience against the failure of a single data center.

[0057] Various embodiments can be implemented in a wide range of operating environments, which in some cases may include one or more user computers or computing devices that can be used to operate any number of applications. User or client devices may include any of many general-purpose personal computers, such as desktop or laptop computers running standard operating systems, as well as cellular, wireless, and portable devices capable of running mobile software and supporting numerous networking and messaging protocols. Such systems may also include numerous workstations capable of running a wide variety of commercially available operating systems and other known applications for purposes such as development and database management. These devices may also include other electronic devices, such as dummy terminals, thin clients, game systems, and other devices capable of communicating over a network.

[0058] Most embodiments utilize one of a variety of commercially available protocols, such as TCP / IP, FTP, UPnP, NFS, and CIFS, and leverage at least one network, which would be well known to those skilled in the art, to support communication. The network may be, for example, a local area network, a wide area network, a virtual private network, the Internet, an intranet, an extranet, a public switched telephone network, an infrared network, a wireless network, or any combination thereof. In embodiments utilizing a web server, the web server may invoke any kind of server or middle-tier application, including HTTP servers, FTP servers, CGI servers, data servers, Java® servers, and business application servers. The server(s) may also have the ability to execute programs or scripts in response to requests from a user device, such as by running one or more web applications, which can be implemented as one or more scripts or programs written in any programming language such as Java®, C, C#, or C++, or any scripting language such as Perl, Python, or TCL, or a combination thereof. Servers may include, but are not limited to, database servers, including those commercially available from Oracle®, Microsoft®, Sybase®, and IBM®.

[0059] The environment may include the various data stores and other memory and storage media described above. These may reside in various locations, such as storage media, that are local to (and / or resident therein) one or more computers, or remote from any or all computers in the network. In a particular set of embodiments, information may reside in a storage area network (SAN) familiar to those skilled in the art. Similarly, any files necessary to perform functions belonging to a computer, server, or other network device may be stored locally and / or remotely as needed. Where the system includes computerized devices, each such device may include hardware elements that can be electrically connected via a bus, which include, for example, at least one central processing unit (CPU), at least one input device (e.g., mouse, keyboard, controller, touch-sensitive display element, or keypad), and at least one output device (e.g., display device, printer, or speaker). Such a system may also include disk drives, optical storage devices, and solid-state storage devices such as random-access memory (RAM) or read-only memory (ROM), as well as one or more storage devices such as removable media devices, memory cards, and flash cards. Such a device may also include, as described above, a computer-readable storage medium reader, a communication device (e.g., a modem, a network card (wireless or wired), an infrared communication device), and working memory. The computer-readable storage medium reader may be connected to, or configured to receive, storage media for temporarily and / or more permanently containing, storing, transmitting, and retrieving computer-readable information, together with computer-readable storage media representing remote, local, fixed, and / or removable storage devices.

[0060] The system and various devices also typically include several software applications, modules, services, or other elements located within at least one working memory device, including an operating system and application programs such as client applications or web browsers. It should be understood that alternative embodiments may have numerous variations from those described above. For example, custom-made hardware may also be used, and / or certain elements may be implemented in hardware, software (including portable software such as applets), or both. Furthermore, connections to other computing devices, such as network input / output devices, may be used. Storage media and other non-temporary computer-readable media containing code or parts of code may include, but are not limited to, any suitable media known or used in the art, such as volatile and non-volatile, removable and non-removable media implemented in any method or technique for storing information such as computer-readable instructions, data structures, program modules, or other data, including RAM, ROM, EEPROM, flash memory or other memory technologies, CD-ROM, digital versatile disk (DVD) or other optical storage devices, magnetic cassettes, magnetic tapes, magnetic disk storage devices or other magnetic storage devices, or any other media that can be used to store desired information and can be accessed by system devices. Based on the disclosures and teachings provided herein, those skilled in the art will understand other methods and / or techniques for implementing diverse embodiments.

[0061] Figures 12A and 12B show exemplary histograms 1200a and 1200b for predicting the number of peptides successfully manufactured, paired with the results from attempts to manufacture those peptides. Figure 12A is an exemplary histogram showing how often a particular predicted number of peptide manufacturability occurs in the dataset. The x-axis 1220a, labeled “Expected Manufacturing,” can quantify the number of peptides expected to be successfully manufactured, while the y-axis 1210a shows the binned frequency of these predictions across the entire dataset. The dashed vertical line represents the number of peptides actually manufactured and serves as a baseline for the predictions. Figure 12A shows that the number of peptides expected to be successfully manufactured is 16.63, while the actual number of peptides successfully manufactured is 15. Figure 12B is an exemplary histogram showing the distribution of manufacturability probabilities (e.g., p(manufacturability)). The x-axis 1220b represents the probability (0-1) that a peptide can be manufactured, and the y-axis 1210b shows the binned frequencies of peptides with these probabilities. Figure 12B shows the reliability of the predictive model for peptide manufacturability. For example, for peptides with high probability scores, the model is confident that those peptides can be successfully manufactured. In one embodiment, the expected number of peptides to be manufactured is calculated by summing the pairwise products between each frequency and probability. In other words, to calculate the overall expected number of manufacturable peptides for the entire dataset, the product of the probability of manufacturability for each peptide and the number of times that probability occurs can be summed. To calculate the expected number of peptides, each intrinsic probability can be multiplied by the number of peptides assigned to that probability, and then these products over all intrinsic probability values ​​can be summed.

[0062] Figures 13A and 13B show exemplary histograms (1300a and 1300b) illustrating the expected number of peptides to be produced and a comparison between two different methods using the same axis as 1200a. The vertical dashed line in both figures shows the actual number of peptides produced using the novel method, in this case the actual number of peptides is 24. Figure 13A first shows the first method, in which peptides are selected from a peptide pool based solely on criteria other than manufacturability. The selected peptides can then be passed through a manufacturing simulation process, which leads to an expected production of an average of 17.59 peptides. Figure 13B shows the improved method, in which peptides are selected based on a combination of manufacturability scores and other criteria. The expected number produced using the second method is approximately 22.90, of which 24 were produced when peptides were selected using this method.

[0063] Therefore, the specification and drawings should be considered illustrative rather than restrictive. However, it is clear that various modifications and changes can be made thereto without departing from the broader spirit and scope of the invention as stated in the claims.

Claims

1. A computer implementation method, Receiving requests for information related to the manufacturability of peptides, To determine whether the aforementioned peptide is expected to be synthesizable, To determine whether the peptide is expected to be soluble, and The computer implementation method includes determining whether the peptide is synthesizable using a trained machine learning model, at least in part, based on whether the peptide is predicted to be synthesizable and soluble.

2. To determine whether the peptide is expected to pass the quality control check, and The computer implementation method according to claim 1, further comprising determining whether the peptide is synthesizable, soluble, and is expected to pass the quality control checks, at least in part, on the basis of determining whether the peptide is manufacturable.

3. The calculation of the feasibility score of the peptide, wherein the feasibility score indicates the probability of successfully manufacturing the peptide, and The computer implementation method according to claim 1 or 2, further comprising determining whether the peptide is manufacturable using the manufacturability score.

4. The computer implementation method according to any one of the prior claims, wherein the machine learning model is trained at least in part with peptide data, peptide feature data, manufacturer specifications and descriptor data, or manufacturability data.

5. A computer implementation method according to any one of the prior claims, wherein the result of whether the peptide was successfully manufactured is used to further train the machine learning model.

6. The computer implementation method according to any one of the prior claims, wherein the determination of whether the peptide is predicted to be synthesizable and the determination of whether the peptide is predicted to be soluble are performed using models specifically trained for each determination.

7. A computer implementation method according to any one of the prior claims, further comprising selecting the peptide for production based on the determination of whether the peptide is manufacturable.

8. A computing system, Computing device processor, and A memory device including instructions, wherein, when the instructions are executed by the computing device processor, the computing system Receiving requests for information related to the manufacturability of peptides, To determine whether the aforementioned peptide is expected to be synthesizable, To determine whether the peptide is expected to be soluble, and The computing system includes a memory device that enables determining whether the peptide is synthesizable using a trained machine learning model, at least in part, whether the peptide is synthesizable and soluble.

9. When the instruction is executed by the computing device processor, it is directed to the computing system. To determine whether the peptide is expected to pass the quality control check, and The computing system according to claim 8, further enabling the determination of whether the peptide is manufacturable, at least in part, on whether the peptide is synthesizable, soluble, and is expected to pass the quality control checks.

10. When the instruction is executed by the computing device processor, it is directed to the computing system. The calculation of the feasibility score of the peptide, wherein the feasibility score indicates the probability of successfully manufacturing the peptide, and The computing system according to claim 8 or 9, further enabling the determination of whether the peptide is manufacturable using the manufacturability score.

11. The computing system according to any one of claims 8 to 10, wherein the machine learning model is trained at least in part with peptide data, peptide feature data, manufacturer specifications and descriptor data, or manufacturability data.

12. The computing system according to any one of claims 8 to 11, wherein the result of whether the peptide was successfully manufactured is used to further train the machine learning model.

13. The computing system according to any one of claims 8 to 12, wherein the determination of whether the peptide is predicted to be synthesizable and the determination of whether the peptide is predicted to be soluble are performed using models specifically trained for each determination.

14. The computing system according to any one of claims 8 to 13, wherein determining whether the peptide is predicted to be synthesizable and determining whether the peptide is predicted to be soluble are determined as part of using the trained machine learning model to determine whether the peptide is manufacturable.

15. When the instruction is executed by the computing device processor, it is directed to the computing system. Determining the solubility score of the peptide, Determining the synthesis feasibility score of the peptide, and The computing system according to any one of claims 8 to 14, further enabling the prediction of the likelihood of the peptide's manufacturability based at least in part on the solubility score or the synthesizability score.

16. A non-temporary computer-readable medium containing stored instructions, wherein the instructions are executed by a processor. Receiving requests for information related to the manufacturability of peptides, To determine whether the aforementioned peptide is expected to be synthesizable, To determine whether the peptide is expected to be soluble, and The non-temporary computer-readable medium performs the step of determining whether the peptide is manufacturable using a trained machine learning model, at least in part, on whether the peptide is predicted to be synthesizable and soluble.

17. When the aforementioned instruction is executed by the processor, To determine whether the peptide is expected to pass the quality control check, and The non-temporary computer-readable medium according to claim 16, further comprising the step of determining whether the peptide is manufacturable, at least in part, on whether the peptide is synthesizable, soluble, and is expected to pass the quality control checks.

18. When the aforementioned instruction is executed by the processor, Determining the solubility score of the peptide, Determining the synthesis feasibility score of the peptide, and A non-transient computer-readable medium according to claim 16 or 17, further comprising the step of predicting the likelihood of the peptide's manufacturability based at least in part on the solubility score or the synthesizability score.

19. The non-transient computer-readable medium according to any one of claims 16 to 18, wherein the machine learning model is trained at least in part on peptide data, peptide feature data, manufacturer specifications and descriptor data, or manufacturability data.

20. A non-temporary computer-readable medium according to any one of claims 16 to 19, wherein the result of whether the peptide was successfully manufactured is used to further train the machine learning model.

Images