Biological detection systems and methods for setting medically suitable target ranges of attributes using principal component analysis (PCA)

Abstract

Biological detection systems and methods are disclosed for defining medically suitable target ranges of molecular attributes. A trial session dataset is received defining a set of subjects and biological experiment data of a test molecule provided to the set of subjects during a trial session. A principal component analysis (PCA) algorithm inputs such data and outputs principal component(s) defining a biological response to the test molecule and correlated with the test molecular attribute. A machine learning classifier inputs the principal component(s)and classifies a plurality of clusters based thereon. First and second positions of respective first and second clusters are selected to define control and sample datasets, respectively. A difference in direction and / or distance from an origin for the first and second positions is detected, and an effective target range is determined therefrom defining a range of attribute exposure values of the test molecule that has a medically suitable biological response correlated with the test molecular attribute.

Description

10688-W001 -SEC (01017-70131 PC)BIOLOGICAL DETECTION SYSTEMS AND METHODS FOR SETTING MEDICALLY SUITABLE TARGET RANGES OF ATTRIBUTES USING PRINCIPAL COMPONENT ANALYSIS (PCA)FIELD OF DISCLOSURE

[0001] The present disclosure generally relates to biological detection systems and methods, and, more particularly, to biological detection systems and methods for defining medically suitable target ranges of molecular attributes using principal component analysis (PCA).BACKGROUND

[0002] Development and manufacture of pharmaceutical and biotechnology products generally requires the measurement or testing of the efficacy and safety of such products. In addition, measurement or testing of therapeutic products can be important to ensure the quality of a development or manufacturing process, and ultimately the quality of the therapeutic products themselves, for the purpose of meeting quality standards and / or regulatory requirements.

[0003] Known approaches for measurement or testing therapeutic products (such as pharmaceutical and biotechnology products) includes using in vitro techniques. However, such techniques may have limited capabilities to accurately assess complex interdependencies relevant to efficacy and safety for humans. This is because the testing of therapeutic products using laboratory techniques in isolation is limited in its ability to measure or test impact of therapeutic products since specific correlations are generally isolated for experimental testing, including complex interactions with respect to various portions of a subject, such as various portions of the human body. For example, conventional techniques can use in vitro datasets, such as lab plate, petri dish, or test tube type data. Such data derived from highly focused experiments can produce bias as these experiments are not designed to investigate the interactions of component species of complex systems against each other, and may incorrectly estimate the importance of one identified interaction towards the whole system. As such, there is a need for analytical systems that are not subject to bias from investigating specific interactions between directly related variables, and instead aim to identify the relative relatedness of component elements of a complex system to each other.

[0004] For these reasons, there is a need for biological detection systems and methods for defining medically suitable target ranges of molecular attributes using PCA, which can be used to set on molecular attribute specifications and limits for the manufacturing of therapeutic products, and which can more accurately identify subtle or hidden features of complex mixed datasets comprised of measured experimental parameters of therapeutic products, and further allow for accurately correlating the results of in vitro assays with in vivo effects, and for accurately estimating their impact on medical conditions and / or portions of the body (such as organ systems) from a more holistic data perspective.10688-W001 -SEC (01017-70131 PC)SUMMARY

[0005] The biological detection systems and methods described herein implement PCA analysis with molecular data and subject body datasets to determine patterns and filter out data noise for developing safe and effective therapeutic products, such as pharmaceutical and biotechnology products. The biological detection systems and methods are configured and / or implemented to define medically suitable target ranges of molecular attributes. For example, medically suitable target ranges define safe and efficacious ranges of molecular attributes for treating disease, medical conditions, or otherwise ailments.

[0006] In some aspects, the biological detection systems and methods differ from prior art techniques at least because instead of using in vitro data (which provides limited testing efficacy), the PCA related algorithms as described herein are implemented using datasets that define or otherwise represent actual human size biocharacters. That is, the biological detection systems and methods described herein can test, measure, otherwise identify levels of data that match or otherwise express a real human (or other nonhuman mammal) body or conditions thereof, which allows the biological detection systems and methods to discover output for setting safe and efficacious target ranges using PCA Dataset represents actual human (or other nonhuman mammal) in vivo datasets. It should be noted that, additionally, or alternatively, in vitro data may still be used, for example, in combination with in vivo datasets. However, it should be noted that in vitro data may be used, at least in some embodiments, alone without in vivo data. Furthermore, the in vitro data may be enhanced or otherwise correlated with the in vivo data in order to enhance or improve the biological detection systems and methods described herein.

[0007] Contrary to conventional techniques, the biological detection systems and methods described herein use actual response in vivo data based on clinical trials and that represent a complete or near-complete human body or nonhuman mammal. Such biological detection systems and methods can be implemented in real time.

[0008] Further, the datasets, as used by the biological detection systems and methods described herein, may comprise biodiverse datasets from humans or otherwise test subjects (e.g., nonhuman mammals) in different geographic areas, each of which may be related with different races, medical histories, or other biological differences. In this way, the biological detection systems and methods described herein allow for discovery and development of therapeutic products, such as pharmaceutical and biotechnology products that can be discovered and applied across different geographic areas, different population sets, different races (e.g., Asian, Caucasian), different ages, and / or other biodiverse categories.

[0009] In various aspects, the biological detection systems and methods described herein provide efficacy and safety measurements and testing by identification of which attribute(s) of a given molecule could be driving adverse events (e.g., a stomachache, medical condition, or other aliment), immunogenicity, and / or impacts to potency. The biological detection systems and10688-W001 -SEC (01017-70131 PC) methods described herein use PCA together with machine learning clustering to implement dimensionality reduction to determine important dimensions and clusters to eliminate or reduce the attributes of the given molecule in order to reduce the risk of adverse event, medical condition or ailment, immunogenicity, or impact to potency.

[0010] In some aspects, the techniques described herein relate to a biological detection system configured to define medically suitable target ranges of molecular attributes, the biological detection system including: one or more processors; a computer memory communicatively coupled to the one or more processors; a database storing each of: (a) molecular data representative of molecular attributes including at least one of: asparagine deamidation, sialylation, high molecular weight (HMW) species fragments, glycans, acidic, basic, and main peak species, chemical modification species, sequence variants, media components, buffer components, and medical device components, such as at least one of asparagine deamidation, sialylation, high molecular weight (HMW) species, fragments, and glycans; and (b) a subject body dataset defining in vitro or in vivo data of a subject body and one or more in vivo effects of the molecular attributes on the subject body; a principal component analysis (PCA) algorithm stored in the computer memory and configured to receive the molecular data and the subject body dataset as input; a machine learning classifier configured to cluster an output of the PCA algorithm; and computing instructions stored on the computer memory, and that when executed by the one or more processors, cause the one or more processors to: receive a trial session dataset defining a set of subjects and biological experiment data of a test molecule provided to the set of subjects during a trial session, the biological experiment data defining (i) control data of the test molecule, and (ii) sample data of the test molecule defining a plurality of samples having a distribution of amounts of the test molecule provided to the set of subjects during the trial session, wherein the test molecule has a test molecular attribute selected from the molecular attributes stored in the database; input the trial session dataset, the biological experiment data, and the subject body dataset into the PCA algorithm, wherein, for each dataset in the control data and the sample data, the PCA algorithm outputs one or more principal components defining a biological response to the test molecule and correlated with the test molecular attribute, input the one or more principal components into the machine learning classifier, wherein the machine learning classifier outputs a plurality of clusters based on the one or more principal components, wherein each cluster has a position relative to another cluster, select a first position of a first cluster of the plurality of clusters, the first cluster defining a control dataset of the control data or a sample dataset of the sample data; select a second position of a second cluster of the plurality of clusters, the second cluster defining a control dataset of the control data or a sample dataset of the sample data; detect a difference in direction and / or distance from an origin for the first position of the first cluster and the second position of the second cluster, the difference defining a comparative degree of10688-W001 -SEC (01017-70131 PC) impact of the biological response to the test molecule as correlated with the test molecular attribute for the first cluster and the second cluster;; and generate, based on the comparative degree of impact of the biological response, an effective target range defining a range of attribute exposure values of the test molecule that has a medically suitable biological response correlated with the test molecular attribute. In some aspects, the molecular attribute comprises at least one of asparagine deamidation, sialylation, high molecular weight (HMW) species, fragments, or glycans. The test molecule may be a therapeutic product or candidate therapeutic product, the active ingredient thereof, or a laboratory counterpart of a therapeutic product.

[0011] In some aspects, the techniques described herein relate to a biological detection system, wherein an attribute exposure value from the effective target range is selected for creating or updating a therapeutic product, and wherein the attribute exposure value identified as triggering the medically suitable biological response.

[0012] In some aspects, the techniques described herein relate to a biological detection system, wherein the therapeutic product is created or updated to treat a specific condition including at least one of: preventing or reducing adverse effects selected from one or more of pyrexia, fatigue, confusional state, anti-drug binding antibodies (BAbs), nausea, encephalopathy, dysgeusia, neutropenia, neurotoxicity, and / or cytokine release syndrome.

[0013] In some aspects, the techniques described herein relate to a biological detection system, wherein creating or updating the therapeutic product includes reducing or eliminating attributes of the test molecule outside of the range of attribute exposure values.

[0014] In some aspects, the techniques described herein relate to a biological detection system, wherein the effective target range is loaded into a memory of a computing device configured to detect tested amounts of the test molecule during manufacture of lots including a therapeutic product including the test molecule, and wherein the computing device rejects any lot including the therapeutic product falling outside of the effective target range.

[0015] In some aspects, the techniques described herein relate to a biological detection system, wherein: (a) the trial session is a clinical trial, the trial session dataset is a clinical trial dataset, and the set of subjects is a set of human clinical trial subjects as defined by the trial session dataset; (b) the trial session is a reactions based test, the trial session dataset is a reactions based dataset, and the set of subjects is a set of nonhuman mammals or cultured cells or tissues as defined by the reactions based dataset, optionally wherein the cultured cells include immune cells or members of a cell-based potency assay; or (c) the trial session is a trends based test, the trial session dataset is a trends based dataset, and the set of subjects is a set of subjects having attributes measured over time.

[0016] In some aspects, the techniques described herein relate to a biological detection system, wherein a relative proximity estimation defines a difference in direction and / or distance10688-W001 -SEC (01017-70131 PC) from the origin for the first cluster and the second cluster, and wherein the difference correlates to a molecular attribute level of the test molecular attribute and a corresponding impact of the molecular attribute level on the biological response to the test molecule.

[0017] In some aspects, the techniques described herein relate to a biological detection system, wherein the range of attribute exposure values corresponds to a dosing value or range of dosing values that results in the medically suitable biological response.

[0018] In some aspects, the techniques described herein relate to a biological detection method for defining medically suitable target ranges of molecular attributes, the biological detection method including: receiving, by one or more processors, a trial session dataset defining a set of subjects and biological experiment data of a test molecule provided to the set of subjects during a trial session, wherein a database storing each of: (a) molecular data representative of molecular attributes including at least one of: asparagine deamidation, sialylation, high molecular weight (HMW) species fragments, glycans, acidic, basic, and main peak species, chemical modification species, sequence variants, media components, buffer components, and medical device components, such as at least one of asparagine deamidation, sialylation, high molecular weight (HMW) species, fragments, and glycans; and (b) a subject body dataset defining in vitro or in vivo data of a subject body and one or more in vivo effects of the molecular attributes on the subject body, and wherein the biological experiment data defines (i) control data of the test molecule, and (ii) sample data of the test molecule defining a plurality of samples having a distribution of amounts of the test molecule provided to the set of subjects during the trial session, wherein the test molecule has a test molecular attribute selected from the molecular attributes stored in the database; inputting, by the one or more processors, the trial session dataset, the biological experiment data, and the subject body dataset into a principal component analysis (PCA) algorithm, wherein the PCA algorithm is configured to receive the molecular data and the subject body dataset as input, and wherein, for each dataset in the control data and the sample data, the PCA algorithm outputs one or more principal components defining a biological response to the test molecule and correlated with the test molecular attribute, inputting, by the one or more processors, the one or more principal components into a machine learning classifier, wherein the machine learning classifier is configured to cluster an output of the PCA algorithm, and wherein the machine learning classifier outputs a plurality of clusters based on the one or more principal components, wherein each cluster has a position relative to another cluster, selecting, by the one or more processors, a first position of a first cluster of the plurality of clusters, the first cluster defining a control dataset of the control data or a sample dataset of the sample data; selecting, by the one or more processors, a second position of a second cluster of the plurality of clusters, the second cluster defining a control dataset of the control data or a sample dataset of the sample data; detecting, by the one or more processors, a difference in direction and / or distance from an origin for the10688-W001 -SEC (01017-70131 PC) first position of the first cluster and the second position of the second cluster, the difference defining a comparative degree of impact of the biological response to the test molecule as correlated with the test molecular attribute for the first cluster and the second cluster; and generating, by the one or more processors and based on the comparative degree of impact of the biological response, an effective target range defining a range of attribute exposure values of the test molecule that has a medically suitable biological response correlated with the test molecular attribute. In some aspects, the molecular attributes include at least one of asparagine deamidation, sialylation, high molecular weight (HMW) species, fragments, or glycans.

[0019] In some aspects, the techniques described herein relate to a biological detection method, wherein an attribute exposure value from the effective target range is selected for creating or updating a therapeutic product, and wherein the attribute exposure value identified as triggering the medically suitable biological response.

[0020] In some aspects, the techniques described herein relate to a biological detection method, wherein the therapeutic product is created or updated to treat or prevent a disease or medical condition including at least one of: preventing or reducing adverse effects selected from one or more of pyrexia, fatigue, confusional state, anti-drug binding antibodies (BAbs), nausea, encephalopathy, dysgeusia, neutropenia, neurotoxicity, and / or cytokine release syndrome.

[0021] In some aspects, the techniques described herein relate to a biological detection method, wherein creating or updating the therapeutic product includes reducing or eliminating attributes of the test molecule outside of the range of attribute exposure values.

[0022] In some aspects, the techniques described herein relate to a biological detection method, wherein the effective target range is loaded into a memory of a computing device configured to detect tested amounts of the test molecule during manufacture of lots including a therapeutic product including the test molecule, and wherein the computing device rejects any lot including the therapeutic product falling outside of the effective target range.

[0023] In some aspects, the techniques described herein relate to a biological detection method, wherein: (a) the trial session is a clinical trial, the trial session dataset is a clinical trial dataset, and the set of subjects is a set of human clinical trial subjects as defined by the trial session dataset; (b) the trial session is a reactions based test, the trial session dataset is a reactions based dataset, and the set of subjects is a set of nonhuman mammals or cultured cells or tissues as defined by the reactions based dataset, optionally wherein the cultured cells include immune cells or members of a cell-based potency assay; or (c) the trial session is a trends based test, the trial session dataset is a trends based dataset, and the set of subjects is a set of subjects having attributes measured over time.

[0024] In some aspects, the techniques described herein relate to a biological detection method, wherein a relative proximity estimation defines a difference in direction and / or distance10688-W001 -SEC (01017-70131 PC) from the origin for the first cluster and the second cluster, and wherein the difference correlates to a molecular attribute level of the test molecular attribute and a corresponding impact of the molecular attribute level on the biological response to the test molecule.

[0025] In some aspects, the techniques described herein relate to a biological detection method, wherein the range of attribute exposure values corresponds to a dosing value or range of dosing values that results in the medically suitable biological response.

[0026] In some aspects, the techniques described herein relate to a non-transitory, tangible computer readable medium storing computing instructions for defining medically suitable target ranges of molecular attributes, the computing instructions, when executed by one or more processors, causes the one or more processors to: receive a trial session dataset defining a set of subjects and biological experiment data of a test molecule provided to the set of subjects during a trial session, wherein a database storing each of: (a) molecular data representative of molecular attributes including at least one of: asparagine deamidation, sialylation, high molecular weight (HMW) species fragments, glycans, acidic, basic, and main peak species, chemical modification species, sequence variants, media components, buffer components, and medical device components, such as at least one of asparagine deamidation, sialylation, high molecular weight (HMW) species, fragments, and glycans; and (b) a subject body dataset defining in vitro or in vivo data of a subject body and one or more in vivo effects of the molecular attributes on the subject body, and wherein the biological experiment data defines (i) control data of the test molecule, and (ii) sample data of the test molecule defining a plurality of samples having a distribution of amounts of the test molecule provided to the set of subjects during the trial session, wherein the test molecule has a test molecular attribute selected from the molecular attributes stored in the database; input the trial session dataset, the biological experiment data, and the subject body dataset into a PCA algorithm, wherein a principal component analysis (PCA) algorithm is configured to receive the molecular data and the subject body dataset as input, and wherein, for each dataset in the control data and the sample data, the PCA algorithm outputs one or more principal components defining a biological response to the test molecule and correlated with the test molecular attribute, input the one or more principal components into a machine learning classifier, wherein the machine learning classifier is configured to cluster an output of the PCA algorithm, and wherein the machine learning classifier outputs a plurality of clusters based on the one or more principal components, wherein each cluster has a position relative to another cluster, select a first position of a first cluster of the plurality of clusters, the first cluster defining a control dataset of the control data or a sample dataset of the sample data; select a second position of a second cluster of the plurality of clusters, the second cluster defining a control dataset of the control data or a sample dataset of the sample data; detect a difference in direction and / or distance from an origin for the first position of the first cluster and the second position of the second cluster, the difference defining10688-W001 -SEC (01017-70131 PC) a comparative degree of impact of the biological response to the test molecule as correlated with the test molecular attribute for the first cluster and the second cluster; and generate, based on the comparative degree of impact of the biological response, an effective target range defining a range of attribute exposure values of the test molecule that has a medically suitable biological response correlated with the test molecular attribute. In some aspects, the molecular attribute comprises at least one of asparagine deamidation, sialylation, high molecular weight (HMW) species, fragments, or glycans.

[0027] In some aspects, the techniques described herein relate to a non-transitory, tangible computer readable medium, wherein an attribute exposure value from the effective target range is selected for creating or updating a therapeutic product, and wherein the attribute exposure value identified as triggering the medically suitable biological response.

[0028] In some aspects, the techniques described herein relate to a non-transitory, tangible computer readable medium, wherein the therapeutic product is created or updated to treat a specific condition including at least one of: preventing or reducing adverse effects selected from one or more of pyrexia, fatigue, confusional state, anti-drug binding antibodies (BAbs), nausea, encephalopathy, dysgeusia, neutropenia, neurotoxicity, and / or cytokine release syndrome.

[0029] In some aspects, the techniques described herein relate to a non-transitory, tangible computer readable medium, wherein creating or updating the therapeutic product includes reducing or eliminating attributes of the test molecule outside of the range of attribute exposure values.

[0030] In some aspects, the techniques described herein relate to a non-transitory, tangible computer readable medium, wherein the effective target range is loaded into a memory of a computing device configured to detect tested amounts of the test molecule during manufacture of lots including a therapeutic product including the test molecule, and wherein the computing device rejects any lot including the therapeutic product falling outside of the effective target range.

[0031] In some aspects, the techniques described herein relate to a non-transitory, tangible computer readable medium, wherein: (a) the trial session is a clinical trial, the trial session dataset is a clinical trial dataset, and the set of subjects is a set of human clinical trial subjects as defined by the trial session dataset; (b) the trial session is a reactions based test, the trial session dataset is a reactions based dataset, and the set of subjects is a set of nonhuman mammals or cultured cells or tissues as defined by the reactions based dataset, optionally wherein the cultured cells include immune cells or members of a cell-based potency assay; or (c) the trial session is a trends based test, the trial session dataset is a trends based dataset, and the set of subjects is a set of subjects having attributes measured over time.10688-W001 -SEC (01017-70131 PC)

[0032] In some aspects, the techniques described herein relate to a non-transitory, tangible computer readable medium, wherein a relative proximity estimation defines a difference in direction and / or distance from the origin for the first cluster and the second cluster, and wherein the difference correlates to a molecular attribute level of the test molecular attribute and a corresponding impact of the molecular attribute level on the biological response to the test molecule.

[0033] In some aspects, the techniques described herein relate to a non-transitory, tangible computer readable medium, wherein the range of attribute exposure values corresponds to a dosing value or range of dosing values that results in the medically suitable biological response.

[0034] In accordance with the above, and with the disclosure herein, the present disclosure includes improvements in computer functionality or in improvements to other technologies at least because the claims recite, e.g., computing device implementing a PCA algorithm and a machine learning classifier to generate a transformed dataset with fewer dimensions (e.g., fewer PCs following implementation of the PCA algorithm), which provides a reduction of the original dataset. The transformed dataset, in turn, reduces computational expense when implemented by the underlying computing device as described herein. Further, error rate(s) are reduced by implementing the PCA analysis described herein, thereby eliminating the need to apply test correction(s) to data of a higher dimension for testing therapeutic products, such as pharmaceutical and biotechnology products.

[0035] Further the present disclosure includes element(s) that apply or use computing technology to affect a particular treatment or prophylaxis for a disease or medical condition. For example, the biological detection systems and methods described herein use PCA together with machine learning clustering to implement dimensionality reduction on to determine important dimensions and clusters to eliminate or reduce the attributes of the given molecule in order to reduce an adverse event or otherwise medical condition or ailment, or immunogenicity, or an impact on potency.

[0036] The present disclosure includes specific features other than what is well-understood, routine, conventional activity in the field, or adding unconventional steps that confine the claim to a particular useful application, e.g., including providing biological detection systems and methods for defining medically suitable target ranges of molecular attributes using PCA.

[0037] Advantages will become more apparent to those of ordinary skill in the art from the following description of the preferred embodiments which have been shown and described by way of illustration. As will be realized, the present embodiments may be capable of other and different embodiments, and their details are capable of modification in various respects. Accordingly, the drawings and description are to be regarded as illustrative in nature and not as restrictive.10688-W001 -SEC (01017-70131 PC)BRIEF DESCRIPTION OF THE DRAWINGS

[0038] The Figures described below depict various aspects of the system and methods disclosed therein. It should be understood that each Figure depicts an embodiment of a particular aspect of the disclosed system and methods, and that each of the Figures is intended to accord with a possible embodiment thereof. Further, wherever possible, the following description refers to the reference numerals included in the following Figures, in which features depicted in multiple Figures are designated with consistent reference numerals.

[0039] There are shown in the drawings arrangements which are presently discussed, it being understood, however, that the present embodiments are not limited to the precise arrangements and instrumentalities shown, wherein:

[0040] Figure 1 illustrates an example biological detection system configured to define medically suitable target ranges of molecular attributes, in accordance with various embodiments disclosed herein.

[0041] Figure 2 illustrates an example flowchart of a biological detection method for defining medically suitable target ranges of molecular attributes, in accordance with various embodiments disclosed herein.

[0042] Figure 3 illustrates an example diagram regarding an example implementation of the biological detection method as described for Figure 2 involving a use case for Clinical Data and Asparagine Deamidation of Therapeutic Protein 1 , a bispecific T-cell engaged molecule, in accordance with various embodiments disclosed herein.

[0043] Figure 4 illustrates a diagram regarding an example implementation of the biological detection method as described for Figure 2 involving a use case for Cytokine Release Data and Sialylation of Therapeutic Protein 2, a human lgG1 monoclonal antibody, in accordance with various embodiments disclosed herein.

[0044] Figure 5 illustrates a diagram regarding an example implementation of the biological detection method as described for Figure 2 involving a use case for Cytokine Release Data and High Molecular Weight Species of Therapeutic Protein 3, a Human IgG Monoclonal Antibody Fragment, in accordance with various embodiments disclosed herein.

[0045] Figure 6 illustrates a diagram regarding an example implementation of the biological detection method as described for Figure 2 involving a use case for Cytokine Release Data and Pre- and Post- Infusion Samples of Therapeutic Protein 4, a Human lgG2 / lgG4 Monoclonal Antibody, in accordance with various embodiments disclosed herein.

[0046] Figure 7 illustrates a diagram regarding an example implementation of the biological detection method as described for Figure 2 involving a use case for Potency Data and a Glycan10688-W001 -SEC (01017-70131 PC)Species of a Therapeutic Protein 6, a Human IgG 1 Monoclonal Antibody, in accordance with various embodiments disclosed herein.

[0047] Figure 8 illustrates a diagram regarding an example implementation of the biological detection method as described for Figure 2 involving a use case for Potency Data and Fragments of a Therapeutic Protein 5, a Human Antigen Binding Protein, in accordance with various embodiments disclosed herein.

[0048] The Figures depict preferred embodiments for purposes of illustration only. Alternative embodiments of the systems and methods illustrated herein may be employed without departing from the principles of the invention described herein.DETAILED DESCRIPTION

[0049] The biological detection systems and methods described herein implement PCA analysis to assess clinical trial data, lots with different levels of glycosylation, potency data, and cell-based immunogenicity prediction assay data (including cytokine secretion, T-cell proliferation, or upregulation of cell surface markers) for the determination of the efficacy and safety of therapeutic products such as pharmaceutical and biotechnology products. Various studies have been conducted using biological detection systems and methods described herein, where PCA analysis has been implemented for or on each of, by way of non-limiting example, clinical data and asparagine deamidation of a therapeutic bispecific T-cell engager molecule; cytokine release data and sialylation of a therapeutic human IgG 1 monoclonal antibody; cytokine release data and high molecular weight species of a therapeutic single chain fragment of a human IgG monoclonal antibody; cytokine release data and pre- and post- infusion samples of a therapeutic human lgG2 / lgG4 monoclonal antibody; potency data and a glycan species of a therapeutic human IgG 1 monoclonal antibody; and potency data and fragments of a therapeutic human antigen binding protein.

[0050] The results from the PCA analysis implemented by the biological detection systems and methods described herein provide detection or otherwise insight into the evaluation of drug molecule attributes. For example, for clinical impact of attributes (CIA) data (See WO 2022 / 132982), the output of the systems and methods herein based on the PCA analysis provides clarity when determining whether adverse events are associated with such drug molecule attributes. As a further example, for cell-based assays the output of the systems and methods herein based on the PCA analysis identifies whether cytokine activity is, or is related to a response to specific attributes. As a still further example, for lots with different levels of glycosylation the output of the systems and methods herein based on the PCA analysis can show how glycan forms effects potency and at what levels.

[0051] As described further herein, the biological detection systems and methods of the present disclosure provide an innovative approach by allowing modeling or definition of attribute levels that are safe with a desired level of potency, selecting candidate molecules with desired10688-W001 -SEC (01017-70131 PC) properties, driving drug product manufacturing, and / or setting Quality Target Product Profile (QTTP) target levels for manufacture of therapeutic products such as pharmaceutical and biotechnology products. Examples of therapeutic products include therapeutic proteins, such as canonical monoclonal antibodies, multi-specific antigen binding proteins (such as bispecific antibodies, tri-specific antibodies, and bispecific T-cell engager molecules), fusion protein, ligand traps, growth factors, muteins, as well as peptides, nucleic acids (such as siRNA and antisence oligonucleotides), and small molecules.

[0052] Figure 1 illustrates an example biological detection system 100 configured to define medically suitable target ranges of molecular attributes, in accordance with various embodiments disclosed herein. In the embodiment of Figure 1 , biological detection system 100 may comprise a computing device 102. Computing device 102 may comprise one or more processors, such as processor 110. Computing device 102 may further comprise a memory 108 communicatively coupled to processor 110 via bus 107, where processor 110 may access memory 108 to load algorithms and / or data therefrom for analysis and / or execution. For example, memory 108 may store PCA algorithm 112 which may comprise computing instructions programmed in a computing language such as Java, Go, C#, C++, Python, or the like. PCA algorithm 112 may be programmed or otherwise configured to receive data 103 (e.g., comprising molecular data and a subject body dataset) as input.

[0053] Further, as shown, memory 108 may store a machine learning classifier 114, which may comprise a machine learning model trained to cluster output of the PCA algorithm. The machine learning model may have been trained with a machine learning classification algorithm, which may include by way of non-limiting example, any of K-means, BIRCH, Gaussian Mixture, and / or DBSCAN and OPTICS algorithm(s).

[0054] Still further, computing device 102 may comprise a display screen 104. Display screen 104 may be configured to render various graphics and / or data as described herein, including any of the graphics or information of the diagrams shown for Figures 3-8 herein. Such graphics and / or data may comprise, by way of non-limiting example, clusters of datasets, including a control dataset and a sample dataset, and respective positions and distances between the clusters to visualize test molecules as correlated with test molecular attributes, for example, as described herein.

[0055] With further reference to Figure 1 , computing device 102 may include input / output (I / O) component 109 for receiving and transmitting data to and from a computer network 120. Input / output (I / O) component 109 may also control display of measurement, identification, classification, or other information as described herein on display screen 104.

[0056] Computing device 102 may receive data 103 (e.g., data or datasets) via computer network 120. For example, processor 110 may access database 132 by initiating a request over10688-W001-SEC (01017-70131 PC) computer network 120 to access data 103. Such data 103 may comprise molecular data, subject body datasets, trial sessions datasets, or other information and / or graphics as described or depicted herein. In some aspects, data 103 may be stored in a database 132, which may comprise, by way of non-limiting example a SQL database (e.g., the ORACLE database, DB2 database, or the like) or a NoSQL based database (e.g., the MongoDB database). As shown for Figure 1 , data 103 may be accessed by computing device 102 via computer network 120. When received, data 103 may be loaded into memory 108 for access by processor 110. For example, processor 110 may access data 103 for use as input into PCA algorithm 112 and / or machine learning classifier 114.

[0057] In various aspect, database 132 may store molecular data representative of molecular attributes comprising at least one of: asparagine deamidation, sialylation, high molecular weight (HMW) species, fragments, glycans, acidic, basic, and main peak species, chemical modification species, sequence variants, media components, buffer components, or medical device components. In some aspects, database 132 may store molecular data representative of molecular attributes comprising at least one of: asparagine deamidation, sialylation, high molecular weight (HMW) species, fragments, and glycans. Asparagine deamidation forms a succinimide intermediate, which may then be converted to aspartic or isoaspartic acid, impacting the primary structure of a therapeutic protein. Sialylation levels are a glycoform can potentially contribute to the anti-inflammatory mechanism of action of the molecule (See, e.g., Cobb, BA (2020). “The history of IgG glycosylation and where we are now.” Glycobiology 30(4): 202-213). High molecular weight (HMW) species refer to complexes that comprise more than one molecule of the therapeutic protein (e.g., dimers or trimers thereof), and can be associated with a loss of potency, impaired stability, or immunogenicity in some therapeutic proteins. Fragments of therapeutic proteins can be associated with a loss of potency, impaired stability, or immunogenicity in some therapeutic proteins. Glycans, including canonical N-glycans, as well as O-glycans, can impact the pharmacokinetics, potency, and effector function of therapeutic proteins. The concentration of a main peak species or the acidic or basic peak species may impact the overall efficacy of a drug product outside of target specifications (and the risk of an impact is greater when an accurate model of impact is not available). Chemical modifications in functionally-relevant regions such as binding domains can impact factors such as the pharmacokinetics, potency, efficacy, effector function, and target specificity of a molecule. Sequence variants may also effect factors such as the pharmacokinetics by changing target specifications, as well as potency, efficacy, effector function, and specificity of a molecule.Media components and buffer components of a drug product could cause chemical modifications by catalysis under stress conditions. Medical device components could impact dataset features such as elevated injection site reactions, pain, and soreness in subjects. Further, database 132 may store a subject body dataset defining in vitro or in vivo data of a10688-W001 -SEC (01017-70131 PC) subject body and one or more in vivo effects of the molecular attributes on the subject body. By way of example, such in vivo effects may include one or more of release of cytokines in response to the attribute indicating one or more types of immunological response, such as a pro-inflammatory response, macrophage recruitment response. As an additional example, molecular attributes may cause the cells to which they were introduced to die. As a further example, molecular attributes may induce proliferation of immune cells (to which the molecular attributes were introduced) into other types of immune cells. A “subject body” refers to an organism, or a biological subsystem of an organism which is under treatment using a therapeutic molecule comprising one or more molecular attributes, which in some aspects is the test molecule, or is under treatment with a suitable control therefor. The subject body dataset comprises parameters that refer to or are informative of in vivo effects on the subject body. The subject body dataset may comprise human data, e.g., of human clinical patients. However, it should be noted that the subject body dataset may also comprise nonhuman data, such as a mammal dataset. For example, a nonhuman mammal dataset may comprise data associated with a mammal such as a dog or a nonhuman primate, for example a cynomolgus monkey. Such data for nonhuman mammals may be useful for toxicology studies or analysis, for example, using the biological detection systems and methods as described herein.

[0058] In various aspects, therapeutic products 140 may be created or updated based on an effective target range defining a range of attribute exposure values of a test molecule that has a medically suitable biological response correlated with a test molecular attribute as described herein. It is noted that a medically suitable biological response refers to acceptable levels of adverse events, and / or acceptable levels of therapeutic efficacy.

[0059] Figure 2 illustrates an example flowchart of a biological detection method 200 for defining medically suitable target ranges of molecular attributes, in accordance with various embodiments disclosed herein. Acceptable target ranges of molecular attributes may include drug product ranges within a release specification, or may include molecular attribute ranges at end-of-shelf, or may include molecular attribute ranges that have been determined to not be associated with adverse events by CIA analysis (See WO2022 / 132982). At block 210, method 200 comprises receiving or otherwise accessing, by one or more processors (e.g., processor 110), a trial session dataset defining a set of subjects (e.g., patients) and biological experiment data of a test molecule provided to the set of subjects during a trial session.

[0060] In various aspects, the biological experiment data can define control data of the test molecule. Such control data may comprise one or more of, by way of non-limiting example, LPS Positive Control data, lgG2 Aggerate control data, and the like, for example, as shown for control dataset 401c of Figure 4 or elsewhere herein.10688-W001 -SEC (01017-70131 PC)

[0061] Additionally, or alternatively, the biological experiment data can define sample data of the test molecule defining a plurality of samples having a distribution of amounts of the test molecule provided to the set of subjects during the trial session. Such sample data of a test molecule may comprise sialic acid 28.2%, sialic acid 1 .8%, and the like, for example as shown for sample dataset 401 s1 of Figure 4 or elsewhere herein. A test molecule may include a test molecular attribute selected from the molecular attributes stored in the database.

[0062] The trail session data may comprise one or more types of data including, by way of non-limiting example, clinical data from patients, reactions of immune cells or tissues from people or animals to biotherapeutic attributes, engineered therapeutic proteins reacting in a potency assay ( / .e. mAbs with different levels of glycans), trends in attribute behavior over time (e.g., reversibility of HMW in serum or PBS), or the like.

[0063] Still further, the trial session dataset may vary based on the data it comprises, and the type of subjects involved in obtaining the dataset. For example, in some aspects, the trial session may comprise a clinical trial and the trial session dataset is a clinical trial dataset where the set of subjects is a set of human clinical trial subjects as defined by the trial session dataset. Additionally, or alternatively, in some aspects the trial session may comprise a reactions-based test and the trial session dataset is a reactions-based dataset where the set of subjects is a set of nonhuman mammals or cultured cells or tissues as defined by the reactions-based dataset. In some cases, the cultured cells may comprise immune cells. However, it should be noted that such nonhuman mammal datasets are not limited to immune cell related data. Still further, additionally, or alternatively, in some aspects the trial session may comprise a trends-based test and the trial session dataset is a trends-based dataset, where the set of subjects is a set of subjects having attributes measured over time.

[0064] At block 220, method 200 comprises inputting, by the one or more processors (e.g., processor 110), the trial session dataset, biological experiment data, and a subject body dataset into a principal component analysis (PCA) algorithm (e.g., PCA algorithm 112). In some aspects, the biological experiment data and a subject body dataset may be received from a database (e.g., database 132 over computer network 120). In such aspects, the database may store such data, where molecular data may comprise or otherwise be representative of molecular attributes comprising at least one of: asparagine deamidation, sialylation, high molecular weight (HMW) species fragments, glycans, acidic, basic, and main peak species, chemical modification species, sequence variants, media components, buffer components, or medical device components. In such aspects, the database may store such data, where molecular data may comprise or otherwise be representative of molecular attributes comprising at least one of: asparagine deamidation, sialylation, high molecular weight (HMW) species, fragments, and glycans. Still further, the subject body dataset may define in vitro or in vivo data of a subject body and one or more in vivo effects of the molecular attributes on the subject body. The subject10688-W001 -SEC (01017-70131 PC) body dataset may comprise human data or nonhuman mammal data, such as a dog or a nonhuman primate, e.g., a cynomolgus monkey. Data for nonhuman mammals can be used for toxicology studies or analysis in accordance with the systems and methods described herein.

[0065] With further reference to block 220 of Figure 2, the PCA algorithm (e.g., PCA algorithm 112) is configured to receive the molecular data and the subject body dataset as input. When provided the input, the PCA algorithm analyzes the molecular data and the subject body dataset and outputs one or more principal components defining a biological response to the test molecule and correlated with the test molecular attribute. The PCA algorithm (e.g., PCA algorithm 112) is implemented to reduce the dimensionality of data, showing patterns and relationships. Such implementation can detect or identify correlation between biotherapeutic attributes and safety and efficacy. In particular, the PCA algorithm detects feature(s) in the molecular data and the subject body dataset as covariant and potentially correlated, or likely not related. Said another way, the PCA algorithm implements principal component analysis to discover principal components. More generally, Principal Component Analysis (PCA) is a dimensionality reduction technique used to transform high-dimensional data into a smaller set of uncorrelated variables, known as principal components, to assist with the identification and visualization of patterns and relationships. The PCA algorithm, implemented for example by processor 110, reduces data complexity by geometrically projecting data of the molecular data and the subject body dataset onto lower dimensions called principal components (PCs), and by targeting the best summary of the data, and therefore PCs, by using a limited number of PCs. In some aspects, a first PC can be chosen to minimize the total difference in direction and / or distance from an origin (e.g., an origin, such as a center, of a PC vector or, additionally or alternatively, a 2D plane of clustered variables having respective positions) for the data and their projection onto the PC. Any second (or subsequent) PCs are selected similarly, with the additional requirement that they be uncorrelated with all previous PCs. A transformed dataset with fewer dimensions (e.g., fewer PCs following implementation of the PCA algorithm) provides a summary or simplification of the original dataset. The transformed dataset, in turn, reduces computational expense when implemented by a computing device (e.g., computing device 102) as described herein. Further, error rate(s), may also be reduced by implementing PCA thereby eliminating the need to apply test correction(s) to data of a higher dimension when testing each feature for association with a particular outcome.

[0066] In some aspects, the first principal component (PC) may be the only principal component. In other embodiments, a the PCA algorithm may comprise multiple PCs. In some aspects, the PCA algorithm may generate a PCA model that comprises the PCA components. The PCA model may be accessed by the PCA algorithm, when executed by processor 110, in order to define a biological response to the test molecule and correlated with the test molecular attribute as described herein.10688-W001 -SEC (01017-70131 PC)

[0067] At block 230, method 200 comprises inputting, by the one or more processors (e.g., processor 110), the one or more principal components into a machine learning classifier. The machine learning classifier is trained or otherwise configured to cluster an output of the PCA algorithm. For example, the machine learning classifier can be trained or configured to output a plurality of clusters (e.g., first cluster 402 and second cluster 452 as shown for Figure 4 or elsewhere herein) based on the one or more principal components where each cluster has a position relative to another cluster (e.g., first position 404 and second position 454 as shown for Figure 4 or elsewhere herein).

[0068] In various aspects, the machine learning classifier comprises an artificial intelligence (Al) based model trained with at least one Al algorithm. Machine learning model(s), such as the machine learning classifier (e.g., machine learning classifier 114) as described herein for some embodiments, may be created and trained based upon example data (e.g., training data) inputs or data (which may be termed “features” and “labels”) in order to make valid and reliable predictions or classifications for new inputs, such as testing level or production level data or inputs. In supervised machine learning, a machine learning program operating on a server, computing device, or otherwise processor(s), may be provided with example inputs (e.g., “features”) and their associated, or observed, outputs (e.g., “labels”) in order for the machine learning program or algorithm to determine or discover rules, relationships, patterns, or otherwise machine learning “models” that map such inputs (e.g., “features”) to the outputs (e.g., labels), for example, by determining and / or assigning weights or other metrics to the model across its various feature categories. Such rules, relationships, or otherwise models may then be provided subsequent inputs in order for the model, executing on the server, computing device, or otherwise processor(s), to predict, based on the discovered rules, relationships, or model, an expected output.

[0069] In unsupervised machine learning, the server, computing device, or otherwise processor(s), may be required to find its own structure in unlabeled example inputs, where, for example multiple training iterations are executed by the server, computing device, or otherwise processor(s) to train multiple generations of models until a satisfactory model, e.g., a model that provides sufficient prediction or classification accuracy when given test level or production level data or inputs, is generated.

[0070] Supervised learning and / or unsupervised machine learning may also comprise retraining, relearning, or otherwise updating models with new, or different, information, which may include information received, ingested, generated, or otherwise used over time. The disclosures herein may use one or both of such supervised or unsupervised machine learning techniques.10688-W001 -SEC (01017-70131 PC)

[0071] Training of the machine learning classifier involves provided as training data output of the PCA algorithm. The machine learning classifier is a type of machine learning model. The machine learning classifier (e.g., machine learning model) may be trained using a supervised or unsupervised machine learning program or algorithm. The machine learning program or algorithm may employ a neural network, which may be a deep learning neural network. The machine learning programs or algorithms may also include regression analysis, support vector machine (SVM) analysis, decision tree analysis, random forest analysis, K-Nearest neighbor analysis, naive Bayes analysis, clustering, reinforcement learning, and / or other machine learning algorithms and / or techniques. In some embodiments, the artificial intelligence and / or machine learning based algorithms may be included as a library or package executed on a server. For example, libraries may include the TENSORFLOW based library, the PYTORCH library, and / or the SCIKIT-LEARN Python library. Additionally, or alternatively, a machine learning algorithm used to train the machine learning classifier may include by way of nonlimiting example, any of K-means, BIRCH, Gaussian Mixture, and / or DBSCAN and OPTICS algorithm(s).

[0072] Training the machine learning classifier (e.g., machine learning classifier 114) adjusting weights of the model of the machine learning classifier to learn patterns in existing data (such as identifying features, such as molecular attributes of a given molecule associated with one or more principal components or their related values as described herein) in order to facilitate making predictions, classifications, or identification for subsequent data, such as using the machine learning classifier on new principal component(s) as output by the PCA algorithm after receiving new data (e.g., data 103) of new trial session dataset, biological experiment data, and subject body dataset in order to cluster an output of the PCA algorithm as described herein.

[0073] With further reference to Figure 2, at block 240, method 200 comprises selecting, by the one or more processors (e.g., processor 110), a first position (e.g., a center of cluster and / or other portion of a cluster) of a first cluster of the plurality of clusters. The first cluster defines a control dataset of the control data or a sample dataset of the sample data. In some aspects, the first position may be a center of a cluster in a first plot of a 2D plane. In other aspects, the first position may be arbitrarily set or selected in a first plot of the 2D plane. In such aspects, the first position may correspond to a position arbitrarily set or otherwise selected relative to a 2D plane using one of the variables, i.e., a first variable, determined by the PCA algorithm and / or PCA model. The first position may be a position in a first plot in the 2D plane. The remaining variables (as determined by the PCA algorithm and / or PCA model) may then be clustered relative to an origin and the first position of the first variable. More generally, implementing the PCA algorithm, and / or using the PCA model, a same set of data could generate a different looking plot in the 2D plane (for example a mirror image of the first plot) but with the overall same clustering of points and directionality and magnitude of the vectors.10688-W001 -SEC (01017-70131 PC)

[0074] It will be appreciated that a difference between the first position of the first cluster and the second position of the second cluster may be detected as a difference in direction or angle of the vectors extending from an origin (e.g., a center or otherwise origin of 2D plane) to each cluster (indicating the extent to which the clusters vary in terms of their principal components), or a difference in their distance from the origin (reflecting the magnitude or intensity of each cluster's influence relative to the dataset as a whole). These differences may define the comparative degree of the biological response to the test molecule as correlated with the test molecular attribute for the first cluster and second cluster.

[0075] At block 250, method 200 comprises selecting, by the one or more processors (e.g., processor 110), a second position (e.g., a center of cluster and / or other portion of a cluster) of a second cluster of the plurality of clusters. The second cluster defines a control dataset of the control data or a sample dataset of the sample data. In some aspects, the second position may be a center of a cluster in a second plot of the 2D plane. In other aspects, the second position may be arbitrarily set or selected in a second plot of the 2D plane. In some aspects, the second position can be determined by the relative relatedness of a second variable (e.g., of the PCA model and / or as determined by the PCA algorithm) to that first variable. In such aspects, the second position can be set by the magnitude of a first vector between the origin and the first variable, and the magnitude of a second vector between the origin and the second variable, along with the directionality (angle between the two vectors) of the two vectors (i.e., the first vector and the second vector) relative to each other.

[0076] At block 260, method 200 comprises detecting, by the one or more processors (e.g., processor 110), a difference in direction and / or distance from an origin for the first position of the first cluster and the second position of the second cluster. The difference defines a comparative degree of impact of the biological response to the test molecule as correlated with the test molecular attribute for the first cluster and the second cluster. In some aspects, a relative proximity estimation defines a difference in direction and / or distance from an origin (e.g., an origin of a PC vector or a 2D plane of clustered variables having respective positions) for the first cluster and the second cluster. Additionally, or alternatively, proximity estimation may define the difference in direction and / or distance from an origin (e.g., an origin of a PC vector or a 2D plane of clustered variables having respective positions) for the first position of the first cluster and the second position of the second cluster. In either case, the difference correlates to a molecular attribute level of the test molecular attribute and a corresponding impact, effect, or otherwise influence of the molecular attribute level on the biological response to the test molecule. The relative proximity estimation is based on proximities of points in clusters, and relative proximities of clusters to other clusters for defining a feature space where the differences between points and clusters correlate to attributes, efficacy, and / or effects of a molecule or molecular attributes on a given sample or control dataset.10688-W001 -SEC (01017-70131 PC)

[0077] Said another way, co-clustering or offset clustering of points can be used to inform or determine similar or differential trends in the complex data. Additionally, proximity of different attribute levels (e.g., as defined by the relative proximity estimation) within various these clusters detect correlation or no correlation which indicates whether an attribute level is favorable or not. For example, attribute levels having a high relative proximity estimation, e.g., close to a strong antigenic positive control ( / .e. LPS or PHA), could be considered as not medically suitable, whereas attribute levels close to a potency assay might be desirable as they cause greater potency ( / .e. level of afucosylation of a mAb and impact on effector function activity).

[0078] With further reference to Figure 2, at block 270, method 200 comprises generating, by the one or more processors (e.g., a processor 110) and based on the comparative degree of impact of the biological response, an effective target range defining a range of attribute exposure values of the test molecule that has a medically suitable biological response correlated with the test molecular attribute. It will be appreciated that if a molecular attribute, or attribute exposure value thereof, co-clusters with an attribute known not to present a risk, then there is a strong indication that the attribute under investigation is likely also not a risk factor. In some aspects, the range of attribute exposure values may correspond to a dosing value or range of dosing values that results in the medically suitable biological response. The dosing value may be a single number or value.

[0079] At block 280, method 200 optionally comprises creating or updating a therapeutic product based on an attribute exposure value selected from the effective target range. The attribute exposure value may be identified as a value triggering the medically suitable biological response. In some aspects, an attribute exposure value from the effective target range is selected for creating or updating a therapeutic product. The attribute exposure value is identified as triggering the medically suitable biological response. By way of example, the release specification for the therapeutic product may be updated based on the attribute exposure value selected from the effective target range. When manufacturing the therapeutic product, batches that are outside of the release specification may be considered not suitable for release, and may be rejected, or may be modified to bring them within the specification.

[0080] In some aspects, therapeutic product is created or updated to treat or prevent a disease or medical condition comprising at least one of: preventing or reducing adverse effects selected from one or more of pyrexia, fatigue, confusional state, anti-drug binding antibodies (BAbs), nausea, encephalopathy, dysgeusia, neutropenia, neurotoxicity, and / or cytokine release syndrome, for example, as shown and described, for example, for Figure 3 herein.

[0081] With further reference to Figure 2, at block 290, method 200 optionally comprises reducing or eliminating attributes of the test molecule outside of the range of attribute exposure values. In this way, the biological detection systems and methods disclosed herein provide a10688-W001 -SEC (01017-70131 PC) particular treatment or prophylaxis for a disease or medical condition. For example, the biological detection systems and methods described herein use PCA together with machine learning clustering to implement dimensionality reduction on to determine important dimensions and clusters to eliminate or reduce the attributes of the given molecule in order to reduce an adverse event or otherwise medical condition or ailment, or immunogenicity, or impact to potency.

[0082] In some aspects, the effective target range can be loaded into a memory of a computing device configured to detect tested amounts of the test molecule during manufacture of lots comprising a therapeutic product comprising the test molecule. A computing device (e.g., computing device102) or server (e.g., server 130) can then reject any lot comprising the therapeutic product falling outside of the effective target range. Said another way, the attribute range can be used for lot inspection or release during manufacturing, where a nonconforming lot gets rejected if it fails to fall within a specified attribute range.

[0083] Example Implementations.

[0084] Figures 3-8, as described further herein, relate to example implementations of the biological detection systems and methods (e.g., system 100 and method 200). The examples are provided to illustrate the biological detection systems and methods, including the use of the PCA algorithm for generating an evidence-based guideline for safe specification levels which can later be confirmed and officially set (e.g., as used for manufacturing or testing therapeutic products for regulatory purposes) using a combination of other approaches including the PCA data and / or other data described herein. The biological detection systems and methods may be used to identify attribute levels and adverse events or immunogenicity or impacts to potency that correlate and that may be outside of one or more acceptable medical or pharmaceutical metrics. For example, if an adverse event such as headache clusters with a high molecular weight (HMW) species dosing, such clustering might indicate that a HMW dose was high enough to elicit an impermissible level of the adverse event (e.g., a headache). Re-running the analysis with a lower max allowable HMW level (by scrubbing the highest-level exposures of HMW) may allow for an improved detection or understanding of the HMW level at which the HMW dose no longer clusters with headache.

[0085] The biological detection systems and methods can detect or otherwise confirm that an attribute of interest clusters with other attributes that are not considered risks, and as such indicates that the attribute of interest is likely or unlikely to be a risk factor, as the case may be. In this way, the biological detection systems and methods can generate evidence-based context on complicated covariant features of a complex system to analyze and understand whether features of interest behave similarly or differently than other features of interest relative to some10688-W001 -SEC (01017-70131 PC) factor (e.g. attributes compared to a given adverse event or adverse events or immunogenicity or impact on potency).

[0086] Example implementation regarding Clinical Data and Asparagine Deamidation of a Therapeutic Protein 1 , a Therapeutic Half-Life Extended Bispecific T Cell Engager Molecule.

[0087] Figure 3 illustrates an example diagram 300 regarding an example implementation of the biological detection method as described for Figure 2 involving a use case for clinical data and asparagine deamidation of Therapeutic Protein 1 , a therapeutic half-life extended bispecific T cell engager molecule, in accordance with various embodiments disclosed herein. Figure 3 illustrates data associated with a Therapeutic Protein 1 showing CIA clinical data as used in a PCA analysis, e.g., implementing the PCA algorithm or otherwise PCA model, for example, as described herein. For example, Figure 3 shows loadings of Therapeutic Protein 1 for a N353 deamidation with dosage for 10 adverse events. The PCA analysis illustrates that exposure of patients to Therapeutic Protein 1 N353 deamidation is correlated with the dose administered but is not correlated with 10 of the most frequent and / or otherwise most importantly detectable adverse events.

[0088] As shown for Figure 3, diagram 300 illustrates that the biological detection systems and methods (e.g., system 100 and method 200) herein, by implementing the PCA algorithm or otherwise PCA analysis, can detect exposure of patients to asparagine deamidation is correlated with a dose 360 administered to the patients, but is not correlated with ten example adverse events, which include a list of most frequent or otherwise important adverse events, as shown for Figure 3, including pyrexia 31 1 , fatigue 312, confusional state 313, anti-drug binding antibodies (BABs) 314, nausea 315, encephalopathy 316, dysgeusia 317, neutropenia 318, neurotoxicity 319, and / or cytokine release syndrome 320. The noncorrelation between these adverse effects and the dosage (e.g., dose 360) of the administered pharmaceutical and biotechnology product, which includes a test molecule, e.g., a therapeutic half-life extended bispecific T cell engager molecule.

[0089] In the example of Figure 3, and with reference to method 200 of Figure 2, processor 110 can receive trial session dataset defining a set of subjects (e.g., human subjects) and biological experiment data (e.g., control and sample data) of a test molecule (e.g., the therapeutic half-life extended bispecific T cell engager molecule) provided to the set of subjects during a trial session.

[0090] Processor 1 10 may then input trial session dataset, the biological experiment data, and the subject body dataset into the PCA algorithm (e.g., PCA algorithm 112). Each of the datasets and data may comprise clinical data regarding asparagine deamidation of a therapeutic half-life extended bispecific T cell engager molecule. For each dataset in the control and sample10688-W001 -SEC (01017-70131 PC) data of the biological experiment data, the PCA algorithm outputs one or more principal components (e.g., a first principal component 300p1 and a second principal component 300p2) defining a biological response to the test molecule and correlated with a test molecular attribute (e.g., asparagine deamidation) of the test molecule. The first principal component 300p1 and the second principal component 300p2 define principal components having values along the x-y axes depicted for Figure 3, where the values indicate comparative degrees of correlation between or among the sample and control data plotted or mapped in the 2-dimensional space of diagram 300 of Figure 3.

[0091] The one or more principal components may be provided, by processor 110, as input to a machine learning classifier (e.g., machine learning classifier 114) trained to output a plurality of clusters (e.g. first cluster 302 and second cluster 352) having a position relative to one another.

[0092] Processor 110 selects a first position (e.g., first position 304) of the first cluster (e.g. first cluster 302). The first cluster can define a control dataset of the control data or a sample dataset of the sample data. In the example of Figure 3, first cluster 302 illustrates a sample dataset showing adverse events (e.g., adverse events 311-320).

[0093] Processor 110 selects a second position (e.g., second position 354) of the second cluster (e.g. second cluster 352). The second cluster can define a control dataset of the control data or a sample dataset of the sample data. In the example of Figure 3, second cluster 352 illustrates a sample dataset showing asparagine deamidation at dose 360.

[0094] Processor 110 can then detect a difference in distance from an origin and / or direction (e.g., difference 370) between the first position of the first cluster and the second position of the second cluster. In the example of Figure 3, the origin may comprise a PC vector including the first principal component 300p1 and the second principal component 300p2. Additionally, or alternatively, the origin may comprise an origin of a 2D plane (e.g., the 2D plane as shown for diagram 300), which may comprise a zeroed (0,0) X-Y position in the 2D plane, and which may be used to define clustered variables having respective positions for the first position of the first cluster and the second position of the second cluster. The difference defines a comparative degree of impact of the biological response to the test molecule as correlated (e.g., in this case uncorrelated) with the test molecular attribute for the first cluster and the second cluster. As shown in the example of Figure 3, the test molecule is uncorrelated with the adverse events (e.g., adverse events 311-320) therefore defining a low comparative impact of the test molecule causing adverse events 311 -320.

[0095] Processor 110 can generate, based on the comparative degree of impact of the biological response, an effective target range defining a range of attribute exposure values of the test molecule that has a medically suitable biological response correlated with the test10688-W001-SEC (01017-70131 PC) molecular attribute. In the example of Figure 3, the effective target range may comprise a range of attribute exposure values, for example 0-10% or 0-5% HMW species that corresponds to a dosing value or range of dosing values of a pharmaceutical and biotechnology product that results in the medically suitable biological response, for example, as illustrated for Figure 3.

[0096] Example implementation regarding Cytokine Release Data and Sialylation of Therapeutic Protein 2, a Therapeutic Human lgG1 Monoclonal Antibody.

[0097] Figure 4 illustrates a diagram 400 regarding an example implementation of the biological detection method as described for Figure 2 involving a use case for cytokine release data and sialylation of Therapeutic Protein 2, a therapeutic human IgG 1 monoclonal antibody, in accordance with various embodiments disclosed herein.

[0098] As shown for Figure 4, diagram 400 illustrates that the biological detection systems and methods (e.g., system 100 and method 200) herein, by implementing the PCA algorithm or otherwise PCA analysis, can detect that different levels of sialic acid (e.g., sample data) do not impact, influence, or otherwise effect cytokine release (e.g., a type of adverse event typically related with cytokine release syndrome where a subject’s immune system overreacts, releasing a large amount of cytokines).

[0099] In the example of Figure 4, datasets of samples with 28.2% sialic acid and 0.3% sialic acid analyzed to determine or detect immune modulatory effect of sialic acid. As shown, it was found that the sialic acid sialylation state of these samples does not change the potential risk of immune activation of cytokine release syndrome for the cell line assay. The example of Figure 4 shows a PCA implementation that detects a non-correlation between sialic acid and immune activation.

[0100] Legend 401 labels several control and sample datasets for Figure 4, where such datasets define data seven days following administration of the control or sample. These include control datasets 401c comprising control data (e.g., cluster 480 for PHA Positive control data, first cluster 402 for lgG2 aggerate control data, cluster 481 for Therapeutic Protein 2 aggregate control data), sample datasets 401 s1 comprising sialic acid sample data (e.g., second cluster 452 for Therapeutic Protein 2 (28.2% Sialic acid) control data, cluster 482 for Therapeutic Protein 2 (1 .8% Sialic acid) control data ), and sample datasets 401 s2 comprising sialic and an activator mixture sample data regarding activating cytokine release data (e.g., cluster 483 for Activator Mixture plus Therapeutic Protein 2 (28.2% Sialic acid) data, cluster 484 for Activator Mixture plus Therapeutic Protein 2 (1 .8% Sialic acid) data).

[0101] In the example of Figure 4, and with reference to method 200 of Figure 2, processor 110 can receive trial session dataset defining a set of subjects (e.g., human subjects) and biological experiment data (e.g., control and sample data as shown and described for Legend10688-W001 -SEC (01017-70131 PC)401 ) of a test molecule (e.g., the therapeutic human IgG 1 monoclonal antibody) provided to the set of subjects during a trial session.

[0102] Processor 110 may then input trial session dataset, the biological experiment data, and the subject body dataset into the PCA algorithm (e.g., PCA algorithm 112). Each of the datasets and data may comprise cytokine release data regarding sialylation of a therapeutic human IgG 1 monoclonal antibody. For each dataset in the control and sample data of the biological experiment data, the PCA algorithm outputs one or more principal components (e.g., a first principal component 400p1 and a second principal component 400p2) defining a biological response to the test molecule and correlated with a test molecular attribute (e.g., cytokine release) of the test molecule. The first principal component 400p1 and the second principal component 400p2 define principal components having values along the x-y axes depicted for Figure 4, where the values indicate comparative degrees of correlation between or among the sample and control data plotted or mapped in the 2-dimensional space of diagram 400 of Figure 4.

[0103] The one or more principal components may be provided, by processor 110, as input to a machine learning classifier (e.g., machine learning classifier 114) trained to output a plurality of clusters (e.g. first cluster 402 and second cluster 452) having a position relative to one another. Zoomed view 452z depicts select data of control datasets 401c and of sample datasets 401 s1 demonstrating a strong correlation between the test molecule (as illustrated from the sample datasets 401 s1) and data of the control datasets 401c.

[0104] Processor 110 selects a first position (e.g., first position 404) of the first cluster (e.g. first cluster 402). The first cluster can define a control dataset of the control data or a sample dataset of the sample data. In the example of Figure 4, first cluster 402 illustrates a control dataset as selected from control datasets 401c having control data and showing impact of an lgG2 Aggregate control on a given subjects.

[0105] Processor 110 selects a second position (e.g., second position 454) of the second cluster (e.g. second cluster 452). The second cluster can define a control dataset of the control data or a sample dataset of the sample data. In the example of Figure 4, second cluster 452 illustrates a sample dataset as selected from sample datasets 401 s1 having sample data and showing impact of a 28.2% sialic acid sample on a given subjects.

[0106] Processor 110 can then detect a difference in distance from an origin and / or direction (e.g., difference 470) between the first position of the first cluster and the second position of the second cluster. In the example of Figure 4, the origin may comprise a PC vector including the first principal component 400p1 and the second principal component 400p2. Additionally, or alternatively, the origin may comprise an origin of a 2D plane (e.g., the 2D plane as shown for diagram 400), which may comprise a zeroed (0,0) X-Y position in the 2D plane, and which may10688-W001 -SEC (01017-70131 PC) be used to define clustered variables having respective positions for the first position of the first cluster and the second position of the second cluster. The difference defines a comparative degree of impact of the biological response to the test molecule as correlated, or otherwise as uncorrelated, with the test molecular attribute for the first cluster and the second cluster. As shown in the example of Figure 4, the data of the test molecule (e.g., as shown for first cluster 402) is correlated with the control datasets 401c, including the lgG2 Aggregate control, therefore defining a similar comparative impact of the test molecule with respect to the control thus indicating or detecting that the test molecule has a low impact or effect regarding cytokine release.

[0107] Processor 110 can generate, based on the comparative degree of impact of the biological response, an effective target range defining a range of attribute exposure values of the test molecule that has a medically suitable biological response correlated with the test molecular attribute. In the example of Figure 4, the effective target range may comprise a range of attribute exposure values that corresponds to a dosing value or range of dosing values of a pharmaceutical and biotechnology product that results in the medically suitable biological response, for example, as illustrated for Figure 4. This could include, for example, a pharmaceutical and biotechnology product is measured to have 1 .8% sialic acid to 28.2% sialic acid to cluster with a comparator molecule with medically acceptable immunogenicity, and / or to cluster far from high immunogenicity molecules such as PHA positive control. Clustering as such, for example, as shown for the sample dataset 401 s1 of Figure 4, indicates that indicating that the therapeutic product under analysis with a measured sialylation between 1 .8% and 28.2% is not a strongly immunogenic molecule, and similarly immunogenic to a comparator molecule with medically acceptable immunogenicity. Accordingly, a release specification for the therapeutic molecule may be set based on the range of 1 .8% sialic acid to 28.2% sialic acid.

[0108] Example implementation regarding Cytokine Release Data and High Molecular Weight Species of Therapeutic Protein 3, a Therapeutic Single Chain Fragment of a Human IgG Monoclonal Antibody.

[0109] Figure 5 illustrates a diagram 500 regarding an example implementation of the biological detection method as described for Figure 2 involving a use case for cytokine release data and high molecular weight species of Therapeutic Protein 3, a therapeutic single chain fragment of a human IgG monoclonal antibody, in accordance with various embodiments disclosed herein.

[0110] As shown for Figure 5, diagram 500 illustrates that the biological detection systems and methods (e.g., system 100 and method 200) herein, by implementing the PCA algorithm or otherwise PCA analysis, can detect whether there is an immune response from 2.5ug / mL. In example of Figure 5, a high molecular weight species attribute (e.g., “HMW2”) is analyzed at10688-W001 -SEC (01017-70131 PC)2%, 5% and 10% HMW2 for immune activation (e.g., cytokine release). The biological detection systems and methods as described herein, including the PCA algorithm, were implemented with respect to cytokine secretion data. Figure 5 illustrates that clinical results indicate that no (or at most a low) immune response is detected for 2.5ug / mL. Further, 2% and 5% HMW2 does not show a higher response than a monomer (e.g., monomer cluster 532), though 10% (e.g., ten percent cluster 544) HMW2 exhibits a low-level immune response.

[0111] Legend 501 labels several control and sample datasets for Figure 5, where such datasets define data seven days following administration of the control or sample. These include control datasets 501c comprising control data (e.g., cluster 581 for PHA late phase positive control data, 10E4 aggerate control cluster 522 for 10E4 aggerate control data, first cluster 502 for Therapeutic Protein 3 Aggregate control data, monomer cluster 532 for Therapeutic Protein 3 monomer data) and sample datasets 501s comprising HMW2 sample data (e.g., cluster 582 for Therapeutic Protein 3 at 2% HMW2 data, cluster 583 for Therapeutic Protein 3 at 5% HMW2 data, second cluster 552 for Therapeutic Protein 3 at 10% HMW2 data). A monomer is used as a comparison against the test molecule, so it is listed with the control datasets 501c.

[0112] In the example of Figure 5, and with reference to method 200 of Figure 2, processor 110 can receive trial session dataset defining a set of subjects (e.g., human subjects) and biological experiment data (e.g., control and sample data as shown and described for Legend 501) of a test molecule (e.g., a therapeutic single chain fragment of a human igg monoclonal antibody) provided to the set of subjects during a trial session.

[0113] Processor 110 may then input trial session dataset, the biological experiment data, and the subject body dataset into the PCA algorithm (e.g., PCA algorithm 112). Each of the datasets and data may comprise cytokine release data regarding high molecular weight species of a therapeutic single chain fragment of a human IgG monoclonal antibody. For each dataset in the control and sample data of the biological experiment data, the PCA algorithm outputs one or more principal components (e.g., a first principal component 500p1 and a second principal component 500p2) defining a biological response to the test molecule and correlated with a test molecular attribute (e.g., cytokine release) of the test molecule. The first principal component 500p1 and the second principal component 500p2 define principal components having values along the x-y axes depicted for Figure 5, where the values indicate comparative degrees of correlation between or among the sample and control data plotted or mapped in the 2- dimensional space of diagram 500 of Figure 5.

[0114] The one or more principal components may be provided, by processor 110, as input to a machine learning classifier (e.g., machine learning classifier 114) trained to output a plurality of clusters (e.g. f6 and second cluster 552) having a position relative to one another. Zoomed view 552z depicts select data of control datasets 501c and of sample datasets 501s10688-W001 -SEC (01017-70131 PC) demonstrating a strong correlation between the test molecule (as illustrated from the sample datasets 501s) and data of the control datasets 501c, including the monomer.

[0115] Processor 110 selects a first position (e.g., first position 504) of the first cluster (e.g. first cluster 502, related to a cluster of data for 2% HMW2). The first cluster can define a control dataset of the control data or a sample dataset of the sample data. In the example of Figure 5, first cluster 502 illustrates a control dataset as selected from control datasets 501c having control data and showing impact of the HMW2 Aggregate control on a given subjects.

[0116] Processor 110 selects a second position (e.g., second position 554) of the second cluster (e.g. second cluster 552). The second cluster can define a control dataset of the control data or a sample dataset of the sample data. In the example of Figure 5, second cluster 552 illustrates a sample dataset as selected from sample datasets 501 s having sample data and showing impact of a 5% HMW2 sample on a given subjects.

[0117] Processor 110 can then detect a difference in distance from an origin and / or direction (e.g., difference 570) between the first position of the first cluster and the second position of the second cluster. In the example of Figure 5, the origin may comprise a PC vector including the first principal component 500p1 and the second principal component 500p2. Additionally, or alternatively, the origin may comprise an origin of a 2D plane (e.g., the 2D plane as shown for diagram 500), which may comprise a zeroed (0,0) X-Y position in the 2D plane, and which may be used to define clustered variables having respective positions for the first position of the first cluster and the second position of the second cluster. The difference defines a comparative degree of impact of the biological response to the test molecule as correlated, or otherwise as uncorrelated, with the test molecular attribute for the first cluster and the second cluster. As shown in the example of Figure 5, the data of the test molecule (e.g., as shown for first cluster 502) is correlated with the control datasets 501c, including the HMW2 Aggregate control, therefore defining a similar comparative impact of the test molecule with respect to the control thus indicating or detecting that the test molecule has a low impact or effect regarding cytokine release.

[0118] Processor 110 can generate, based on the comparative degree of impact of the biological response, an effective target range defining a range of attribute exposure values of the test molecule that has a medically suitable biological response correlated with the test molecular attribute. In the example of Figure 5, the effective target range may comprise a range of attribute exposure values that corresponds to a dosing value or range of dosing values of a pharmaceutical and biotechnology product that results in the medically suitable biological response, for example, as illustrated for Figure 5. This could include, for example, a pharmaceutical and biotechnology product including a 2% to 10% HMW2, for example, as shown for the sample dataset 501s of Figure 5.10688-W001 -SEC (01017-70131 PC)

[0119] With further reference to Figure 6, processor 110 may detect additional or otherwise different clusters, including by way of non-limiting example clusters of the control datasets 501c, including 10E4 aggerate control cluster 522 and monomer cluster 532, as well as clusters of the sample datasets 501s, including ten percent cluster 544, related to a cluster of data for 10% HMW2. These additional clusters can also be analyzed for respective differences for updating or further defining the effective target range of attribute exposure values of the test molecule that has the medically suitable biological response correlated with the test molecular attribute.

[0120] Example implementation regarding Cytokine Release Data and Pre- and PostInfusion Samples of Therapeutic Protein 4, a Therapeutic Human lgG2 / lgG4 Monoclonal Antibody.

[0121] Figure 6 illustrates a diagram 600 regarding an example implementation of the biological detection method as described for Figure 2 involving a use case for cytokine release data and pre- and post- infusion samples of Therapeutic Protein 4, a therapeutic human lgG2 / lgG4 monoclonal antibody, in accordance with various embodiments disclosed herein.

[0122] As shown for Figure 6, diagram 600 illustrates that the biological detection systems and methods (e.g., system 100 and method 200) herein, by implementing the PCA algorithm or otherwise PCA analysis, can detect whether pre-infusion and post-infusion samples comprising a test molecule of a therapeutic human lgG2 / lgG4 monoclonal antibody induce a higher immune response compared to a given control (e.g., a diluent or otherwise generate lgG2 control). In example of Figure 6, samples of a therapeutic human lgG2 / lgG4 monoclonal antibody (e.g., sample datasets 601 s1) as well as reference drug product (a drug product of the same monoclonal antibody, but produced under different conditions; e.g., sample datasets 601 s2) are analyzed to detect any immune activation (e.g., cytokine release). The biological detection systems and methods as described herein, including the PCA algorithm, were implemented with respect to cytokine secretion data. Figure 6 illustrates that late phase results indicate that no (or at most a low) immune response is detected for the therapeutic human lgG2 / lgG4 monoclonal antibody samples (e.g., sample datasets 601 s1) as well as the reference drug product samples (e.g., sample datasets 601 s2).

[0123] Legend 601 labels several control and sample datasets for Figure 6, where such datasets define data, several days following administration of the control or sample. These include control datasets 601c comprising control data (e.g., cluster 681 for PHA positive control data, cluster 682 for Immunocult positive control data, first cluster 602 for lgG2 aggerate control data, aggregate control cluster 622 for Therapeutic protein 4 aggregate control data, diluent cluster 632 for IV Diluent plus buffer data), sample datasets 601 s1 comprising therapeutic human lgG2 / lgG4 monoclonal antibody sample data (e.g., cluster 683 for Therapeutic Protein 4 pre-infusion sample data, second cluster 652 for Therapeutic Protein 4 post-infusion sample10688-W001-SEC (01017-70131 PC) data), and sample datasets 601 s2 comprising reference drug product sample data (e.g., cluster 684 for Therapeutic Protein 4 pre-infusion sample data, cluster 642 for Therapeutic Protein 4 post-infusion sample data). A diluent is used as a comparison against the test molecule (therapeutic human lgG2 / lgG4 monoclonal antibody) and the reference drug product, so it is listed with the control datasets 601c, e.g., as diluent cluster 632. Further, in the example of Figure 6, clusters 632, 642, 652, 683, and 684 are completely or at least partially overlapping, such that the clusters, at least in some instances, are depicted superimposed on one another.

[0124] In the example of Figure 6, and with reference to method 200 of Figure 2, processor 110 can receive trial session dataset defining a set of subjects (e.g., human subjects) and biological experiment data (e.g., control and sample data as shown and described for Legend 601 ) of a test molecule (e.g., a therapeutic human lgG2 / lgG4 monoclonal antibody) and the reference drug product provided to the set of subjects during a trial session.

[0125] Processor 110 may then input the trial session dataset, the biological experiment data, and the subject body dataset into the PCA algorithm (e.g., PCA algorithm 112). Each of the datasets and data may comprise cytokine release data and pre- and post- infusion samples of a therapeutic human lgG2 / lgG4 monoclonal antibody. For each dataset in the control and sample data of the biological experiment data, the PCA algorithm outputs one or more principal components (e.g., a first principal component 600p1 and a second principal component 600p2) defining a biological response to the test molecule and correlated with a test molecular attribute (e.g., cytokine release) of the test molecule. The first principal component 600p1 and the second principal component 600p2 define principal components having values along the x-y axes depicted for Figure 6, where the values indicate comparative degrees of correlation between or among the sample and control data plotted or mapped in the 2-dimensional space of diagram600 of Figure 6.

[0126] The one or more principal components may be provided, by processor 110, as input to a machine learning classifier (e.g., machine learning classifier 114) trained to output a plurality of clusters (e.g. first cluster 602 and second cluster 652) having a position relative to one another. Zoomed view 652z depicts select data of control datasets 601c and of sample datasets601 s1 and 601 s2 demonstrating a strong correlation among the test molecule and reference drug product (as illustrated from the sample datasets 601 s1 and 601 s2) and data of the control datasets 601c, including the diluent.

[0127] Processor 110 selects a first position (e.g., first position 604) of the first cluster (e.g. first cluster 602, related to a cluster of data for an lgG2 Aggregate control). The first cluster can define a control dataset of the control data or a sample dataset of the sample data. In the example of Figure 6, first cluster 602 illustrates a control dataset as selected from control10688-W001-SEC (01017-70131 PC) datasets 601c having control data and showing impact of the lgG2 Aggregate control on a given subjects.

[0128] Processor 110 selects a second position (e.g., second position 654) of the second cluster (e.g. second cluster 652). The second cluster can define a control dataset of the control data or a sample dataset of the sample data. In the example of Figure 6, second cluster 652 illustrates a sample dataset as selected from sample datasets 601 s having sample data and showing lack of cluster differentiation between immunogenicity response of the test molecule (e.g., therapeutic human lgG2 / lgG4 monoclonal antibody, post-infusion) pre-infusion and postinfusion.

[0129] Processor 110 can then detect a difference in distance from an origin and / or direction (e.g., difference 670) between the first position of the first cluster and the second position of the second cluster. In the example of Figure 6, the origin may comprise a PC vector including the first principal component 600p1 and the second principal component 600p2. Additionally, or alternatively, the origin may comprise an origin of a 2D plane (e.g., the 2D plane as shown for diagram 600), which may comprise a zeroed (0,0) X-Y position in the 2D plane, and which may be used to define clustered variables having respective positions for the first position of the first cluster and the second position of the second cluster. The difference defines a comparative degree of impact of the biological response to the test molecule as correlated, or otherwise as uncorrelated, with the test molecular attribute for the first cluster and the second cluster. As shown in the example of Figure 6, the data of the test molecule (e.g., as shown for first cluster 602) is correlated with the control datasets 601c, including the lgG2 Aggregate control, therefore defining a similar comparative impact of the test molecule with respect to the control thus indicating or detecting that the test molecule has a low impact or effect regarding cytokine release.

[0130] Processor 110 can generate, based on the comparative degree of impact of the biological response, an effective target range defining a range of attribute exposure values of the test molecule that has a medically suitable biological response correlated with the test molecular attribute. In the example of Figure 6, the effective target range may comprise a range of attribute exposure values that corresponds to a dosing value or range of dosing values of a pharmaceutical and biotechnology product that results in the medically suitable biological response, for example, as illustrated for Figure 6. This could include, for example, a pharmaceutical and biotechnology product including a pre-infusion or post infusion of the test molecule, for example, as shown for the sample dataset 601 s1 of Figure 6.

[0131] With further reference to Figure 6, processor 110 may detect additional or otherwise different clusters, including by way of non-limiting example clusters of the control datasets 601c, including aggregate control cluster 622 and diluent cluster 632, as well as clusters of the sample10688-W001 -SEC (01017-70131 PC) datasets 601 s1 and 601 s2, including reference drug product cluster, related to a cluster of data for reference drug product post-infusion. Additional clusters are shown for control and sample data regarding pre-and post-infusion of the target molecule and reference product drug product. These additional clusters can also be analyzed for respective differences for updating or further defining the effective target range of attribute exposure values of the test molecule that has the medically suitable biological response correlated with the test molecular attribute.

[0132] Example implementation regarding Potency Data and a Glycan Species of Therapeutic Protein 6, a Therapeutic Human lgG1 Monoclonal Antibody.

[0133] Figure 7 illustrates a diagram 700 regarding an example implementation of the biological detection method as described for Figure 2 involving a use case for potency data and a glycan species of Therapeutic Protein 6, a therapeutic human IgG 1 monoclonal antibody, in accordance with various embodiments disclosed herein.

[0134] As shown for Figure 7, diagram 700 illustrates that the biological detection systems and methods (e.g., system 100 and method 200) herein, by implementing the PCA algorithm or otherwise PCA analysis, can detect differences and similarities among clusters of data samples clustered based on glycan attributes (e.g., galactosylated glycans (Gal) or otherwise Gal%, high mannose (HM) or otherwise HM%, and the like) and their influence on therapeutic human IgG 1 monoclonal antibody potency. In the example, in example of Figure 7, the biological detection systems and methods as described herein, including the PCA algorithm, were implemented with respect to potency related data and a glycan species of a therapeutic human IgG 1 monoclonal antibody. Figure 7 illustrates that clones cluster (e.g., clusters 702, 712, 722, and 752 related to datasets 701 d1 , 701 d2, 701 d3, and 701 d4, respectively) based on a given parent pool. Figure 7 also illustrates that attributes Gal, HM, and other glycans have dominant influences on therapeutic human lgG1 monoclonal antibody potency (e.g., FcgR3a potency).

[0135] Legend 701 labels several datasets for Figure 7, where such datasets define data of the various datasets 701 d1 (as depicted for cluster 722), 701 d2 (as depicted for cluster 752), 701 d3 (as depicted for cluster 712), and 701 d4 (as depicted for cluster 702). These include datasets regarding potency data and a glycan species of a therapeutic human IgG 1 monoclonal antibody.

[0136] In the example of Figure 7, and with reference to method 200 of Figure 2, processor 110 can receive trial session dataset defining a set of subjects (e.g., human subjects) and biological experiment data (e.g., data as shown and described for Legend 701 ) of a test molecule (e.g., a therapeutic human IgG 1 monoclonal antibody) provided to the set of subjects during a trial session.

[0137] Processor 1 10 may then input trial session dataset, the biological experiment data, and the subject body dataset into the PCA algorithm (e.g., PCA algorithm 112). Each of the10688-W001 -SEC (01017-70131 PC) datasets and data may comprise potency data and a glycan species regarding therapeutic human IgG 1 monoclonal antibody potency. Any of the datasets may be selected as a control or sample dataset. For each dataset in the control and sample data of the biological experiment data, the PCA algorithm outputs one or more principal components (e.g., a first principal component 700p1 and a second principal component 700p2) defining a biological response to the test molecule and correlated with a test molecular attribute (e.g., galactosylated glycans (Gal) or otherwise Gal%, high mannose (HM) or otherwise HM%, and the like) of the test molecule. The first principal component 700p1 and the second principal component 700p2 define principal components having values along the x-y axes depicted for Figure 7, where the values indicate comparative degrees of correlation between or among the data plotted or mapped in the 2-dimensional space of diagram 700 of Figure 7.

[0138] The one or more principal components may be provided, by processor 110, as input to a machine learning classifier (e.g., machine learning classifier 114) trained to output a plurality of clusters (e.g. first cluster 702 and second cluster 752) having a position relative to one another. As shown, each of first cluster 702 and second cluster 752 depicts select data of datasets 701 d1 and 701 d2 demonstrating correlation and, thus clustering based on a given parent pool.

[0139] Processor 110 selects a first position (e.g., first position 704) of the first cluster (e.g. first cluster 702, related to a cluster of data related to parent 500P). The first cluster can define a control dataset of the control data or a sample dataset of the sample data. In the example of Figure 7, first cluster 702 illustrates a control dataset as selected from control dataset 701 d4 having control data and showing impact of the therapeutic human IgG 1 monoclonal antibody (e.g., FcgR3a) on a given subjects.

[0140] Processor 110 selects a second position (e.g., second position 754) of the second cluster e.g. second cluster 752, related to a cluster of data related to parent 150P). The second cluster can define a control dataset of the control data or a sample dataset of the sample data. In the example of Figure 7, second cluster 752 illustrates a sample dataset as selected from sample datasets 701 d2 having sample data and showing impact of the therapeutic human lgG1 monoclonal antibody (e.g., FcgR3a) on a given subjects.

[0141] Processor 110 can then detect a difference in distance from an origin and / or direction (e.g., difference770) between the first position of the first cluster and the second position of the second cluster. In the example of Figure 7, the origin may comprise a PC vector including the first principal component 700p1 and the second principal component 700p2. Additionally, or alternatively, the origin may comprise an origin of a 2D plane (e.g., the 2D plane as shown for diagram 700), which may comprise a zeroed (0,0) X-Y position in the 2D plane, and which may be used to define clustered variables having respective positions for the first position of the first10688-W001 -SEC (01017-70131 PC) cluster and the second position of the second cluster. The difference defines a comparative degree of impact of the biological response to the test molecule as correlated, or otherwise as uncorrelated, with the test molecular attribute for the first cluster and the second cluster. As shown in the example of Figure 7, the data of the test molecule (e.g., as shown for first cluster 702) is correlated among the datasets 701 d2 and 702d4 (as illustrated by clusters 752 and 702, respectfully), therefore defining a similar comparative impact of the test molecule with for each of the datasets thus indicating or detecting that the test molecule has a similar impact for the two datasets.

[0142] Processor 110 can generate, based on the comparative degree of impact of the biological response, an effective target range defining a range of attribute exposure values of the test molecule that has a medically suitable biological response correlated with the test molecular attribute. In the example of Figure 7, the effective target range may comprise a range of attribute exposure values that corresponds to a dosing value or range of dosing values of a pharmaceutical and biotechnology product that results in the medically suitable biological response, for example, as illustrated for Figure 7. This could include, for example, a pharmaceutical and biotechnology product including a range of values associated with the test molecule, for example, as shown for the sample datasets 701d1-701d4 of Figure 7.

[0143] With further reference to Figure 7, processor 110 may detect additional or otherwise different clusters, including by way of non-limiting example dataset 701 d1 related to cluster 722 and dataset 701d3 related to cluster 712. These additional clusters can also be analyzed for respective differences for updating or further defining the effective target range of attribute exposure values of the test molecule that has the medically suitable biological response correlated with the test molecular attribute.

[0144] Example implementation regarding Potency Data and Fragments of Therapeutic Protein 5, a Therapeutic Human Antigen Binding Protein.

[0145] Figure 8 illustrates a diagram 800 regarding an example implementation of the biological detection method as described for Figure 2 involving a use case for potency data and fragments of Therapeutic Protein 5, a therapeutic human antigen binding protein, in accordance with various embodiments disclosed herein.

[0146] As shown for Figure 8, diagram 800 illustrates that the biological detection systems and methods (e.g., system 100 and method 200) herein, by implementing the PCA algorithm or otherwise PCA analysis, can detect a positive correlation of a Attribute C (a fragment, e.g., cluster 802) with a cell based potency assay “A” (e.g., cluster 852), but does not correlate with a Therapeutic Protein 5 potency binding assay (e.g., cluster 812) or a different cell based potency assay “B” (e.g., cluster 822). Further as shown, other clusters, (e.g., cluster 832, cluster 842, and cluster 862) show Attribute C-spiked sample data, which is also positively correlated with10688-W001 -SEC (01017-70131 PC)Attribute C (e.g., cluster 802). Where positively correlated, the data illustrates potency of a test molecule, e.g., fragments of a therapeutic human antigen binding protein.

[0147] In the example of Figure 8, and with reference to method 200 of Figure 2, processor 110 can receive trial session dataset defining a set of subjects (e.g., human subjects) and biological experiment data (e.g., data as shown and described above) of a test molecule (e.g., a therapeutic human antigen binding protein) provided to the set of subjects during a trial session.

[0148] Processor 110 may then input trial session dataset, the biological experiment data, and the subject body dataset into the PCA algorithm (e.g., PCA algorithm 112). Each of the datasets and data may comprise potency data regarding fragments of a therapeutic human antigen binding protein. For each dataset in the control and sample data of the biological experiment data, the PCA algorithm outputs one or more principal components (e.g., a first principal component 800p1 and a second principal component 800p2) defining a biological response to the test molecule and correlated with a test molecular attribute (e.g., potency) of the test molecule. The first principal component 800p1 and the second principal component 800p2 define principal components having values along the x-y axes depicted for Figure 8, where the values indicate comparative degrees of correlation between or among the data plotted or mapped in the 2-dimensional space of diagram 800 of Figure 8.

[0149] The one or more principal components may be provided, by processor 110, as input to a machine learning classifier (e.g., machine learning classifier 114) trained to output a plurality of clusters (e.g. first cluster 802 and second cluster 852) having a position relative to one another. Diagram 800 depicts select data of datasets from clusters 802 and 852 demonstrating a strong correlation among the test molecule. Data of clusters 832, 842, and 862 indicate a positive correlation among one another, and also with clusters 802 and 852, as each cluster exhibits patterns extending in a same direction (e.g., as shown by corresponding arrows). Data and related clusters 812 and 822, however, are uncorrelated with first cluster 802 and second cluster 852.

[0150] Processor 110 selects a first position (e.g., first position 804) of the first cluster (e.g. first cluster 802, related to a cluster of data for Attribute C). The first cluster can define a control dataset of the control data or a sample dataset of the sample data. In the example of Figure 8, first cluster 802 illustrates a control dataset having control data and showing impact of the Attribute C on a given subjects.

[0151] Processor 110 selects a second position (e.g., second position 854) of the second cluster (e.g. second cluster 852). The second cluster can define a control dataset of the control data or a sample dataset of the sample data. In the example of Figure 8, second cluster 852 illustrates a sample dataset having sample data and showing impact of the test molecule (e.g., fragments of a therapeutic human antigen binding protein) on a given subjects.10688-W001 -SEC (01017-70131 PC)

[0152] Processor 110 can then detect a difference in distance from an origin and / or direction (e.g., difference 870) between the first position of the first cluster and the second position of the second cluster. In the example of Figure 8, the origin may comprise a PC vector including the first principal component 800p1 and the second principal component 800p2. Additionally, or alternatively, the origin may comprise an origin of a 2D plane (e.g., the 2D plane as shown for diagram 800), which may comprise a zeroed (0,0) X-Y position in the 2D plane, and which may be used to define clustered variables having respective positions for the first position of the first cluster and the second position of the second cluster. The difference defines a comparative degree of impact of the biological response to the test molecule as correlated, or otherwise as uncorrelated, with the test molecular attribute for the first cluster and the second cluster. As shown in the example of Figure 8, the data of the test molecule (e.g., as shown for first cluster 802) is correlated with the control dataset, (e.g., shown for cluster 802), including Attribute C, therefore defining a similar comparative impact of the test molecule with respect to the control thus indicating or detecting a similar potency.

[0153] Processor 1 10 can generate, based on the comparative degree of impact of the biological response, an effective target range defining a range of attribute exposure values of the test molecule that has a medically suitable biological response correlated with the test molecular attribute. In the example of Figure 8, the effective target range may comprise a range of attribute exposure values that corresponds to a dosing value or range of dosing values of a pharmaceutical and biotechnology product that results in the medically suitable biological response, for example, as illustrated for Figure 8. This could include, for example, a pharmaceutical and biotechnology product of the test molecule, for example, as shown for cluster 852 of Figure 8.

[0154] With further reference to Figure 8, processor 110 may detect additional or otherwise different clusters, including by way of non-limiting example clusters 812, 822, 832, 842, and 862. These additional clusters can also be analyzed for respective differences for updating or further defining the effective target range of attribute exposure values of the test molecule that has the medically suitable biological response correlated with the test molecular attribute. For example, cluster 812 with position 814 for the Therapeutic Protein 5 potency binding assay has a greater comparative difference 871 when compared with difference 870. Similarly, cluster 822 with position 824 for cell-based potency assay “B” has a greater comparative difference 872 when compared with difference 870. These greater differences indicate a lower degree of impact of the biological response for each of the Therapeutic Protein 5 potency binding assay and the cellbased potency assay “B” when compared to cell-based potency assay “B” as associated with cluster 852.

[0155] Aspects of the Present Disclosure10688-W001 -SEC (01017-70131 PC)

[0156] The following aspects are provided as examples in accordance with the disclosure herein and are not intended to limit the scope of the disclosure.

[0157] Aspect 1 . A biological detection system configured to define medically suitable target ranges of molecular attributes, the biological detection system comprising: one or more processors; a computer memory communicatively coupled to the one or more processors; a database storing each of: (a) molecular data representative of molecular attributes comprising at least one of: asparagine deamidation, sialylation, high molecular weight (HMW) species fragments, glycans, acidic, basic, and main peak species, chemical modification species, sequence variants, media components, buffer components, and medical device components, , such as at least one of asparagine deamidation, sialylation, high molecular weight (HMW) species, fragments, and glycans; and (b) a subject body dataset defining in vitro or in vivo data of a subject body and one or more in vivo effects of the molecular attributes on the subject body; a principal component analysis (PCA) algorithm stored in the computer memory and configured to receive the molecular data and the subject body dataset as input; a machine learning classifier configured to cluster an output of the PCA algorithm; and computing instructions stored on the computer memory, and that when executed by the one or more processors, cause the one or more processors to: receive a trial session dataset defining a set of subjects and biological experiment data of a test molecule provided to the set of subjects during a trial session, the biological experiment data defining (i) control data of the test molecule, and (ii) sample data of the test molecule defining a plurality of samples having a distribution of amounts of the test molecule provided to the set of subjects during the trial session, wherein the test molecule has a test molecular attribute selected from the molecular attributes stored in the database; input the trial session dataset, the biological experiment data, and the subject body dataset into the PCA algorithm, wherein, for each dataset in the control data and the sample data, the PCA algorithm outputs one or more principal components defining a biological response to the test molecule and correlated with the test molecular attribute, input the one or more principal components into the machine learning classifier, wherein the machine learning classifier outputs a plurality of clusters based on the one or more principal components, wherein each cluster has a position relative to another cluster, select a first position of a first cluster of the plurality of clusters, the first cluster defining a control dataset of the control data or a sample dataset of the sample data; select a second position of a second cluster of the plurality of clusters, the second cluster defining a control dataset of the control data or a sample dataset of the sample data; detect a difference in direction and / or distance from an origin for the first position of the first cluster and the second position of the second cluster, the difference defining a comparative degree of impact of the biological response to the test molecule as correlated with the test molecular attribute for the first cluster and the second cluster; and generate, based on the comparative degree of impact of the biological response, an effective target range10688-W001 -SEC (01017-70131 PC) defining a range of attribute exposure values of the test molecule that has a medically suitable biological response correlated with the test molecular attribute. In some aspects, the molecular attributes comprise at least one of asparagine deamidation, sialylation, high molecular weight (HMW) species, fragments, and glycans.

[0158] Aspect 2. The biological detection system of clause 1 , wherein an attribute exposure value from the effective target range is selected for creating or updating a therapeutic product, and wherein the attribute exposure value identified as triggering the medically suitable biological response.

[0159] Aspect 3. The biological detection system of aspect 2, wherein the therapeutic product is created or updated to treat a specific condition comprising at least one of: preventing or reducing adverse effects selected from one or more of pyrexia, fatigue, confusional state, antidrug binding antibodies (BAbs), nausea, encephalopathy, dysgeusia, neutropenia, neurotoxicity, and / or cytokine release syndrome.

[0160] Aspect 4. The biological detection system of aspect 2, wherein creating or updating the therapeutic product comprises reducing or eliminating attributes of the test molecule outside of the range of attribute exposure values.

[0161] Aspect 5. The biological detection system of any one of aspects 1-4, wherein the effective target range is loaded into a memory of a computing device configured to detect tested amounts of the test molecule during manufacture of lots comprising a therapeutic product comprising the test molecule, and wherein the computing device rejects any lot comprising the therapeutic product falling outside of the effective target range.

[0162] Aspect 6. The biological detection system of any one of aspects 1-5, wherein: (a) the trial session is a clinical trial, the trial session dataset is a clinical trial dataset, and the set of subjects is a set of human clinical trial subjects as defined by the trial session dataset; (b) the trial session is a reactions based test, the trial session dataset is a reactions based dataset, and the set of subjects is a set of nonhuman mammals or cultured cells or tissues as defined by the reactions based dataset, optionally wherein the cultured cells comprise immune cells or members of a cell-based potency assay; or (c) the trial session is a trends based test, the trial session dataset is a trends based dataset, and the set of subjects is a set of subjects having attributes measured over time.

[0163] Aspect 7. The biological detection system of any one of aspects 1-6, wherein a relative proximity estimation defines a difference in direction and / or distance from the origin for the first cluster and the second cluster, and wherein the difference correlates to a molecular attribute level of the test molecular attribute and a corresponding impact of the molecular attribute level on the biological response to the test molecule.10688-W001 -SEC (01017-70131 PC)

[0164] Aspect 8. The biological detection system of any one of aspects 1-7, wherein the range of attribute exposure values corresponds to a dosing value or range of dosing values that results in the medically suitable biological response.

[0165] Aspect 9. A biological detection method for defining medically suitable target ranges of molecular attributes, the biological detection method comprising: receiving, by one or more processors, a trial session dataset defining a set of subjects and biological experiment data of a test molecule provided to the set of subjects during a trial session, wherein a database storing each of: (a) molecular data representative of molecular attributes comprising at least one of: asparagine deamidation, sialylation, high molecular weight (HMW) species fragments, glycans, acidic, basic, and main peak species, chemical modification species, sequence variants, media components, buffer components, and medical device components, such as at least one of asparagine deamidation, sialylation, high molecular weight (HMW) species, fragments, and glycans; and (b) a subject body dataset defining in vitro or in vivo data of a subject body and one or more in vivo effects of the molecular attributes on the subject body, and wherein the biological experiment data defines (i) control data of the test molecule, and (ii) sample data of the test molecule defining a plurality of samples having a distribution of amounts of the test molecule provided to the set of subjects during the trial session, wherein the test molecule has a test molecular attribute selected from the molecular attributes stored in the database; inputting, by the one or more processors, the trial session dataset, the biological experiment data, and the subject body dataset into a principal component analysis (PCA) algorithm, wherein the PCA algorithm is configured to receive the molecular data and the subject body dataset as input, and wherein, for each dataset in the control data and the sample data, the PCA algorithm outputs one or more principal components defining a biological response to the test molecule and correlated with the test molecular attribute, inputting, by the one or more processors, the one or more principal components into a machine learning classifier, wherein the machine learning classifier is configured to cluster an output of the PCA algorithm, and wherein the machine learning classifier outputs a plurality of clusters based on the one or more principal components, wherein each cluster has a position relative to another cluster, selecting, by the one or more processors, a first position of a first cluster of the plurality of clusters, the first cluster defining a control dataset of the control data or a sample dataset of the sample data; selecting, by the one or more processors, a second position of a second cluster of the plurality of clusters, the second cluster defining a control dataset of the control data or a sample dataset of the sample data; detecting, by the one or more processors, a difference in direction and / or distance from an origin for the first position of the first cluster and the second position of the second cluster, the difference defining a comparative degree of impact of the biological response to the test molecule as correlated with the test molecular attribute for the first cluster and the second cluster; and generating, by the one or more processors and based on the comparative degree of10688-W001 -SEC (01017-70131 PC) impact of the biological response, an effective target range defining a range of attribute exposure values of the test molecule that has a medically suitable biological response correlated with the test molecular attribute. In some aspects, the molecular attributes comprise at least one of asparagine deamidation, sialylation, high molecular weight (HMW) species, fragments, or glycans.

[0166] Aspect 10. The biological detection method of aspect 9, wherein an attribute exposure value from the effective target range is selected for creating or updating a therapeutic product, and wherein the attribute exposure value identified as triggering the medically suitable biological response.

[0167] Aspect 11 . The biological detection method of aspect 10, wherein the therapeutic product is created or updated to treat or prevent a disease or medical condition comprising at least one of: preventing or reducing adverse effects selected from one or more of pyrexia, fatigue, confusional state, anti-drug binding antibodies (BAbs), nausea, encephalopathy, dysgeusia, neutropenia, neurotoxicity, and / or cytokine release syndrome.

[0168] Aspect 12. The biological detection method of aspect 10, wherein creating or updating the therapeutic product comprises reducing or eliminating attributes of the test molecule outside of the range of attribute exposure values.

[0169] Aspect 13. The biological detection method of any one of aspects 9-12, wherein the effective target range is loaded into a memory of a computing device configured to detect tested amounts of the test molecule during manufacture of lots comprising a therapeutic product comprising the test molecule, and wherein the computing device rejects any lot comprising the therapeutic product falling outside of the effective target range.

[0170] Aspect 14. The biological detection method of any one of aspects 9-13, wherein: (a) the trial session is a clinical trial, the trial session dataset is a clinical trial dataset, and the set of subjects is a set of human clinical trial subjects as defined by the trial session dataset; (b) the trial session is a reactions based test, the trial session dataset is a reactions based dataset, and the set of subjects is a set of nonhuman mammals or cultured cells or tissues as defined by the reactions based dataset, optionally wherein the cultured cells comprise immune cells or members of a cell-based potency assay; or (c) the trial session is a trends based test, the trial session dataset is a trends based dataset, and the set of subjects is a set of subjects having attributes measured over time.

[0171] Aspect 15. The biological detection method of any one of aspects 9-14, wherein a relative proximity estimation defines a difference in direction and / or distance from the origin for the first cluster and the second cluster, and wherein the difference correlates to a molecular attribute level of the test molecular attribute and a corresponding impact of the molecular attribute level on the biological response to the test molecule.10688-W001 -SEC (01017-70131 PC)

[0172] Aspect 16. The biological detection method of any one of aspects 9-15, wherein the range of attribute exposure values corresponds to a dosing value or range of dosing values that results in the medically suitable biological response.

[0173] Aspect 17. A non-transitory, tangible computer readable medium storing computing instructions for defining medically suitable target ranges of molecular attributes, the computing instructions, when executed by one or more processors, causes the one or more processors to: receive a trial session dataset defining a set of subjects and biological experiment data of a test molecule provided to the set of subjects during a trial session, wherein a database storing each of: (a) molecular data representative of molecular attributes comprising at least one of: asparagine deamidation, sialylation, high molecular weight (HMW) species fragments, glycans, acidic, basic, and main peak species, chemical modification species, sequence variants, media components, buffer components, and medical device components, such as at least one of ; and (b) a subject body dataset defining in vitro or in vivo data of a subject body and one or more in vivo effects of the molecular attributes on the subject body, and wherein the biological experiment data defines (i) control data of the test molecule, and (ii) sample data of the test molecule defining a plurality of samples having a distribution of amounts of the test molecule provided to the set of subjects during the trial session, wherein the test molecule has a test molecular attribute selected from the molecular attributes stored in the database; input the trial session dataset, the biological experiment data, and the subject body dataset into a PCA algorithm, wherein a principal component analysis (PCA) algorithm is configured to receive the molecular data and the subject body dataset as input, and wherein, for each dataset in the control data and the sample data, the PCA algorithm outputs one or more principal components defining a biological response to the test molecule and correlated with the test molecular attribute, input the one or more principal components into a machine learning classifier, wherein the machine learning classifier is configured to cluster an output of the PCA algorithm, and wherein the machine learning classifier outputs a plurality of clusters based on the one or more principal components, wherein each cluster has a position relative to another cluster, select a first position of a first cluster of the plurality of clusters, the first cluster defining a control dataset of the control data or a sample dataset of the sample data; select a second position of a second cluster of the plurality of clusters, the second cluster defining a control dataset of the control data or a sample dataset of the sample data; detect a difference in direction and / or distance from an origin for the first position of the first cluster and the second position of the second cluster, the difference defining a comparative degree of impact of the biological response to the test molecule as correlated with the test molecular attribute for the first cluster and the second cluster; and generate, based on the comparative degree of impact of the biological response, an effective target range defining a range of attribute exposure values of the test molecule that has a medically suitable biological response correlated with the test molecular attribute. In some10688-W001-SEC (01017-70131 PC) aspects, the molecular attributes comprise as least one of asparagine deamidation, sialylation, high molecular weight (HMW) species, fragments, or glycans.

[0174] Aspect 18. The non-transitory, tangible computer readable medium of aspect 17, wherein an attribute exposure value from the effective target range is selected for creating or updating a therapeutic product, and wherein the attribute exposure value identified as triggering the medically suitable biological response.

[0175] Aspect 19. The non-transitory, tangible computer readable medium of aspect 18, wherein the therapeutic product is created or updated to treat a specific condition comprising at least one of: preventing or reducing adverse effects selected from one or more of pyrexia, fatigue, confusional state, anti-drug binding antibodies (BAbs), nausea, encephalopathy, dysgeusia, neutropenia, neurotoxicity, and / or cytokine release syndrome.

[0176] Aspect 20. The non-transitory, tangible computer readable medium of aspect 18, wherein creating or updating the therapeutic product comprises reducing or eliminating attributes of the test molecule outside of the range of attribute exposure values.

[0177] Aspect 21 . The non-transitory, tangible computer readable medium of any one of aspects 17-20, wherein the effective target range is loaded into a memory of a computing device configured to detect tested amounts of the test molecule during manufacture of lots comprising a therapeutic product comprising the test molecule, and wherein the computing device rejects any lot comprising the therapeutic product falling outside of the effective target range.

[0178] Aspect 22. The non-transitory, tangible computer readable medium of any one of aspects 17-21 , wherein: (a) the trial session is a clinical trial, the trial session dataset is a clinical trial dataset, and the set of subjects is a set of human clinical trial subjects as defined by the trial session dataset; (b) the trial session is a reactions based test, the trial session dataset is a reactions based dataset, and the set of subjects is a set of nonhuman mammals or cultured cells or tissues as defined by the reactions based dataset, optionally wherein the cultured cells comprise immune cells or members of a cell-based potency assay; or (c) the trial session is a trends based test, the trial session dataset is a trends based dataset, and the set of subjects is a set of subjects having attributes measured over time.

[0179] Aspect 23. The non-transitory, tangible computer readable medium of any one of aspects 17-22, wherein a relative proximity estimation defines a difference in direction and / or distance from the origin for the first cluster and the second cluster, and wherein the difference correlates to a molecular attribute level of the test molecular attribute and a corresponding impact of the molecular attribute level on the biological response to the test molecule.10688-W001 -SEC (01017-70131 PC)

[0180] Aspect 24. The non-transitory, tangible computer readable medium of any one of aspects 17-23, wherein the range of attribute exposure values corresponds to a dosing value or range of dosing values that results in the medically suitable biological response.

[0181] The foregoing aspects of the disclosure are exemplary only and not intended to limit the scope of the disclosure

[0182] Additional Considerations

[0183] Although the disclosure herein sets forth a detailed description of numerous different embodiments, it should be understood that the legal scope of the description is defined by the words of the claims set forth at the end of this patent and equivalents. The detailed description is to be construed as exemplary only and does not describe every possible embodiment since describing every possible embodiment would be impractical. Numerous alternative embodiments may be implemented, using either current technology or technology developed after the filing date of this patent, which would still fall within the scope of the claims.

[0184] The following additional considerations apply to the foregoing discussion. Throughout this specification, plural instances may implement components, operations, or structures described as a single instance. Although individual operations of one or more methods are illustrated and described as separate operations, one or more of the individual operations may be performed concurrently, and nothing requires that the operations be performed in the order illustrated. Structures and functionality presented as separate components in example configurations may be implemented as a combined structure or component. Similarly, structures and functionality presented as a single component may be implemented as separate components. These and other variations, modifications, additions, and improvements fall within the scope of the subject matter herein.

[0185] Additionally, certain embodiments are described herein as including logic or a number of routines, subroutines, applications, or instructions. These may constitute either software (e.g., code embodied on a machine-readable medium or in a transmission signal) or hardware. In hardware, the routines, etc., are tangible units capable of performing certain operations and may be configured or arranged in a certain manner. In example embodiments, one or more computer systems (e.g., a standalone, client or server computer system) or one or more hardware modules of a computer system (e.g., a processor or a group of processors) may be configured by software (e.g., an application or application portion) as a hardware module that operates to perform certain operations as described herein.

[0186] In various embodiments, a hardware module may be implemented mechanically or electronically. For example, a hardware module may comprise dedicated circuitry or logic that is permanently configured (e.g., as a special-purpose processor, such as a field programmable gate array (FPGA) or an application-specific integrated circuit (ASIC)) to perform certain10688-W001 -SEC (01017-70131 PC) operations. A hardware module may also comprise programmable logic or circuitry (e.g., as encompassed within a general-purpose processor or other programmable processor) that is temporarily configured by software to perform certain operations. It will be appreciated that the decision to implement a hardware module mechanically, in dedicated and permanently configured circuitry, or in temporarily configured circuitry (e.g., configured by software) may be driven by cost and time considerations.

[0187] Accordingly, the term “hardware module” should be understood to encompass a tangible entity, be that an entity that is physically constructed, permanently configured (e.g., hardwired), or temporarily configured (e.g., programmed) to operate in a certain manner or to perform certain operations described herein. Considering embodiments in which hardware modules are temporarily configured (e.g., programmed), each of the hardware modules need not be configured or instantiated at any one instance in time. For example, where the hardware modules comprise a general-purpose processor configured using software, the general-purpose processor may be configured as respective different hardware modules at different times. Software may accordingly configure a processor, for example, to constitute a particular hardware module at one instance of time and to constitute a different hardware module at a different instance of time.

[0188] The term “coupled to” used herein does not require a direct coupling or connection, such that two items may be “coupled to” one another through one or more intermediary components or other elements, such as an electronic bus, electrical wiring, mechanical component, or other such indirect connection.

[0189] Hardware modules may provide information to, and receive information from, other hardware modules. Accordingly, the described hardware modules may be regarded as being communicatively coupled. Where multiple of such hardware modules exist contemporaneously, communications may be achieved through signal transmission (e.g., over appropriate circuits and buses) that connect the hardware modules. In embodiments in which multiple hardware modules are configured or instantiated at different times, communications between such hardware modules may be achieved, for example, through the storage and retrieval of information in memory structures to which the multiple hardware modules have access. For example, one hardware module may perform an operation and store the output of that operation in a memory device to which it is communicatively coupled. A further hardware module may then, at a later time, access the memory device to retrieve and process the stored output. Hardware modules may also initiate communications with input or output devices, and may operate on a resource (e.g., a collection of information).

[0190] The various operations of example methods described herein may be performed, at least partially, by one or more processors that are temporarily configured (e.g., by software) or10688-W001-SEC (01017-70131 PC) permanently configured to perform the relevant operations. Whether temporarily or permanently configured, such processors may constitute processor-implemented modules that operate to perform one or more operations or functions. The modules referred to herein may, in some example embodiments, comprise processor-implemented modules.

[0191] Similarly, the methods or routines described herein may be at least partially processor implemented. For example, at least some of the operations of a method may be performed by one or more processors or processor-implemented hardware modules. The performance of certain of the operations may be distributed among the one or more processors, not only residing within a single machine, but deployed across a number of machines. In some example embodiments, the processor or processors may be located in a single location, while in other embodiments the processors may be distributed across a number of locations.

[0192] The performance of certain of the operations may be distributed among the one or more processors, not only residing within a single machine, but deployed across a number of machines. In some example embodiments, the one or more processors or processor- implemented modules may be located in a single geographic location (e.g., within a home environment, an office environment, or a server farm). In other embodiments, the one or more processors or processor-implemented modules may be distributed across a number of geographic locations.

[0193] This detailed description is to be construed as exemplary only and does not describe every possible embodiment, as describing every possible embodiment would be impractical, if not impossible. A person of ordinary skill in the art may implement numerous alternate embodiments, using either current technology or technology developed after the filing date of this application.

[0194] Those of ordinary skill in the art will recognize that a wide variety of modifications, alterations, and combinations can be made with respect to the above-described embodiments without departing from the scope of the invention, and that such modifications, alterations, and combinations are to be viewed as being within the ambit of the inventive concept.

[0195] The patent claims at the end of this patent application are not intended to be construed under 35 U.S.C. § 112(f) unless traditional means-plus-function language is expressly recited, such as “means for” or “step for” language being explicitly recited in the claim(s). The systems and methods described herein are directed to an improvement to computer functionality, and improve the functioning of conventional computers.

Claims

10688-W001 -SEC (01017-70131 PC)What is claimed is:1 . A biological detection system configured to define medically suitable target ranges of molecular attributes, the biological detection system comprising: one or more processors; a computer memory communicatively coupled to the one or more processors; a database storing each of: (a) molecular data representative of molecular attributes comprising at least one of: asparagine deamidation, sialylation, high molecular weight (HMW) species fragments, glycans, acidic, basic, and main peak species, chemical modification species, sequence variants, media components, buffer components, and medical device components, such as at least one of asparagine deamidation, sialylation, high molecular weight (HMW) species, fragments, and glycans; and (b) a subject body dataset defining in vitro or in vivo data of a subject body and one or more in vivo effects of the molecular attributes on the subject body; a principal component analysis (PCA) algorithm stored in the computer memory and configured to receive the molecular data and the subject body dataset as input; a machine learning classifier configured to cluster an output of the PCA algorithm; and computing instructions stored on the computer memory, and that when executed by the one or more processors, cause the one or more processors to: receive a trial session dataset defining a set of subjects and biological experiment data of a test molecule provided to the set of subjects during a trial session, the biological experiment data defining (i) control data of the test molecule, and (ii) sample data of the test molecule defining a plurality of samples having a distribution of amounts of the test molecule provided to the set of subjects during the trial session, wherein the test molecule has a test molecular attribute selected from the molecular attributes stored in the database; input the trial session dataset, the biological experiment data, and the subject body dataset into the PCA algorithm, wherein, for each dataset in the control data and the sample data, the PCA algorithm outputs one or more principal components defining a biological response to the test molecule and correlated with the test molecular attribute, input the one or more principal components into the machine learning classifier, wherein the machine learning classifier outputs a plurality of clusters based on the one or more principal components, wherein each cluster has a position relative to another cluster, select a first position of a first cluster of the plurality of clusters, the first cluster defining a control dataset of the control data or a sample dataset of the sample data;10688-W001 -SEC (01017-70131 PC) select a second position of a second cluster of the plurality of clusters, the second cluster defining a control dataset of the control data or a sample dataset of the sample data; detect a difference in direction and / or distance from an origin for the first position of the first cluster and the second position of the second cluster, the difference defining a comparative degree of impact of the biological response to the test molecule as correlated with the test molecular attribute for the first cluster and the second cluster; and generate, based on the comparative degree of impact of the biological response, an effective target range defining a range of attribute exposure values of the test molecule that has a medically suitable biological response correlated with the test molecular attribute.

2. The biological detection system of claim 1 , wherein an attribute exposure value from the effective target range is selected for creating or updating a therapeutic product, and wherein the attribute exposure value identified as triggering the medically suitable biological response.

3. The biological detection system of claim 2, wherein the therapeutic product is created or updated to treat a specific condition comprising at least one of: preventing or reducing adverse effects selected from one or more of pyrexia, fatigue, confusional state, antidrug binding antibodies (BAbs), nausea, encephalopathy, dysgeusia, neutropenia, neurotoxicity, and / or cytokine release syndrome.

4. The biological detection system of claim 2, wherein creating or updating the therapeutic product comprises reducing or eliminating attributes of the test molecule outside of the range of attribute exposure values.

5. The biological detection system of claim 1 , wherein the effective target range is loaded into a memory of a computing device configured to detect tested amounts of the test molecule during manufacture of lots comprising a therapeutic product comprising the test molecule, and wherein the computing device rejects any lot comprising the therapeutic product falling outside of the effective target range.

6. The biological detection system of claim 1 , wherein: (a) the trial session is a clinical trial, the trial session dataset is a clinical trial dataset, and the set of subjects is a set of human clinical trial subjects as defined by the trial session dataset; (b) the trial session is a reactions based test, the trial session dataset is a reactions based dataset, and the set of subjects is a set of nonhuman mammals or cultured cells or tissues as defined by the reactions10688-W001 -SEC (01017-70131 PC) based dataset, optionally wherein the cultured cells comprise immune cells or members of a cell-based potency assay; or (c) the trial session is a trends based test, the trial session dataset is a trends based dataset, and the set of subjects is a set of subjects having attributes measured over time.

7. The biological detection system of claim 1 , wherein a relative proximity estimation defines a difference in direction and / or distance from the origin for the first cluster and the second cluster, and wherein the difference correlates to a molecular attribute level of the test molecular attribute and a corresponding impact of the molecular attribute level on the biological response to the test molecule.

8. The biological detection system of claim 1 , wherein the range of attribute exposure values corresponds to a dosing value or range of dosing values that results in the medically suitable biological response.

9. A biological detection method for defining medically suitable target ranges of molecular attributes, the biological detection method comprising: receiving, by one or more processors, a trial session dataset defining a set of subjects and biological experiment data of a test molecule provided to the set of subjects during a trial session, wherein a database storing each of: (a) molecular data representative of molecular attributes comprising at least one of: asparagine deamidation, sialylation, high molecular weight (HMW) species fragments, glycans, acidic, basic, and main peak species, chemical modification species, sequence variants, media components, buffer components, and medical device components, such as at least one of asparagine deamidation, sialylation, high molecular weight (HMW) species, fragments, and glycans; and (b) a subject body dataset defining in vitro or in vivo data of a subject body and one or more in vivo effects of the molecular attributes on the subject body, and wherein the biological experiment data defines (i) control data of the test molecule, and (ii) sample data of the test molecule defining a plurality of samples having a distribution of amounts of the test molecule provided to the set of subjects during the trial session, wherein the test molecule has a test molecular attribute selected from the molecular attributes stored in the database; inputting, by the one or more processors, the trial session dataset, the biological experiment data, and the subject body dataset into a principal component analysis (PCA) algorithm, wherein the PCA algorithm is configured to receive the molecular data and the subject body dataset as input, and10688-W001 -SEC (01017-70131 PC) wherein, for each dataset in the control data and the sample data, the PCA algorithm outputs one or more principal components defining a biological response to the test molecule and correlated with the test molecular attribute, inputting, by the one or more processors, the one or more principal components into a machine learning classifier, wherein the machine learning classifier is configured to cluster an output of thePCA algorithm, and wherein the machine learning classifier outputs a plurality of clusters based on the one or more principal components, wherein each cluster has a position relative to another cluster, selecting, by the one or more processors, a first position of a first cluster of the plurality of clusters, the first cluster defining a control dataset of the control data or a sample dataset of the sample data; selecting, by the one or more processors, a second position of a second cluster of the plurality of clusters, the second cluster defining a control dataset of the control data or a sample dataset of the sample data; detecting, by the one or more processors, a difference in direction and / or distance from an origin for the first position of the first cluster and the second position of the second cluster, the difference defining a comparative degree of impact of the biological response to the test molecule as correlated with the test molecular attribute for the first cluster and the second cluster; and generating, by the one or more processors and based on the comparative degree of impact of the biological response, an effective target range defining a range of attribute exposure values of the test molecule that has a medically suitable biological response correlated with the test molecular attribute.

10. The biological detection method of claim 9, wherein an attribute exposure value from the effective target range is selected for creating or updating a therapeutic product, and wherein the attribute exposure value identified as triggering the medically suitable biological response.11 . The biological detection method of claim 10, wherein the therapeutic product is created or updated to treat or prevent a disease or medical condition comprising at least one of: preventing or reducing adverse effects selected from one or more of pyrexia, fatigue, confusional state, anti-drug binding antibodies (BAbs), nausea, encephalopathy, dysgeusia, neutropenia, neurotoxicity, and / or cytokine release syndrome.10688-W001 -SEC (01017-70131 PC)12. The biological detection method of claim 10, wherein creating or updating the therapeutic product comprises reducing or eliminating attributes of the test molecule outside of the range of attribute exposure values.

13. The biological detection method of claim 9, wherein the effective target range is loaded into a memory of a computing device configured to detect tested amounts of the test molecule during manufacture of lots comprising a therapeutic product comprising the test molecule, and wherein the computing device rejects any lot comprising the therapeutic product falling outside of the effective target range.

14. The biological detection method of claim 9, wherein: (a) the trial session is a clinical trial, the trial session dataset is a clinical trial dataset, and the set of subjects is a set of human clinical trial subjects as defined by the trial session dataset; (b) the trial session is a reactions based test, the trial session dataset is a reactions based dataset, and the set of subjects is a set of nonhuman mammals or cultured cells or tissues as defined by the reactions based dataset, optionally wherein the cultured cells comprise immune cells or members of a cell-based potency assay; or (c) the trial session is a trends based test, the trial session dataset is a trends based dataset, and the set of subjects is a set of subjects having attributes measured over time.

15. The biological detection method of claim 9, wherein a relative proximity estimation defines a difference in direction and / or distance from the origin for the first cluster and the second cluster, and wherein the difference correlates to a molecular attribute level of the test molecular attribute and a corresponding impact of the molecular attribute level on the biological response to the test molecule.

16. The biological detection method of claim 9, wherein the range of attribute exposure values corresponds to a dosing value or range of dosing values that results in the medically suitable biological response.

17. A non-transitory, tangible computer readable medium storing computing instructions for defining medically suitable target ranges of molecular attributes, the computing instructions, when executed by one or more processors, causes the one or more processors to: receive a trial session dataset defining a set of subjects and biological experiment data of a test molecule provided to the set of subjects during a trial session, wherein a database storing each of: (a) molecular data representative of molecular attributes comprising at least one of: asparagine deamidation, sialylation, high molecular weight (HMW) species fragments, glycans, acidic, basic, and main peak10688-W001 -SEC (01017-70131 PC) species, chemical modification species, sequence variants, media components, buffer components, and medical device components, such as at least one of asparagine deamidation, sialylation, high molecular weight (HMW) species, fragments, and glycans; and (b) a subject body dataset defining in vitro or in vivo data of a subject body and one or more in vivo effects of the molecular attributes on the subject body, and wherein the biological experiment data defines (i) control data of the test molecule, and (ii) sample data of the test molecule defining a plurality of samples having a distribution of amounts of the test molecule provided to the set of subjects during the trial session, wherein the test molecule has a test molecular attribute selected from the molecular attributes stored in the database; input the trial session dataset, the biological experiment data, and the subject body dataset into a PCA algorithm, wherein a principal component analysis (PCA) algorithm is configured to receive the molecular data and the subject body dataset as input, and wherein, for each dataset in the control data and the sample data, the PCA algorithm outputs one or more principal components defining a biological response to the test molecule and correlated with the test molecular attribute, input the one or more principal components into a machine learning classifier, wherein the machine learning classifier is configured to cluster an output of the PCA algorithm, and wherein the machine learning classifier outputs a plurality of clusters based on the one or more principal components, wherein each cluster has a position relative to another cluster, select a first position of a first cluster of the plurality of clusters, the first cluster defining a control dataset of the control data or a sample dataset of the sample data; select a second position of a second cluster of the plurality of clusters, the second cluster defining a control dataset of the control data or a sample dataset of the sample data; detect a difference in direction and / or distance from an origin for the first position of the first cluster and the second position of the second cluster, the difference defining a comparative degree of impact of the biological response to the test molecule as correlated with the test molecular attribute for the first cluster and the second cluster; and generate, based on the comparative degree of impact of the biological response, an effective target range defining a range of attribute exposure values of the test molecule that has a medically suitable biological response correlated with the test molecular attribute.

18. The non-transitory, tangible computer readable medium of claim 17, wherein an attribute exposure value from the effective target range is selected for creating or updating a10688-W001 -SEC (01017-70131 PC) therapeutic product, and wherein the attribute exposure value identified as triggering the medically suitable biological response.

19. The non-transitory, tangible computer readable medium of claim 18, wherein the therapeutic product is created or updated to treat a specific condition comprising at least one of: preventing or reducing adverse effects selected from one or more of pyrexia, fatigue, confusional state, anti-drug binding antibodies (BAbs), nausea, encephalopathy, dysgeusia, neutropenia, neurotoxicity, and / or cytokine release syndrome.

20. The non-transitory, tangible computer readable medium of claim 18, wherein creating or updating the therapeutic product comprises reducing or eliminating attributes of the test molecule outside of the range of attribute exposure values.21 . The non-transitory, tangible computer readable medium of claim 17, wherein the effective target range is loaded into a memory of a computing device configured to detect tested amounts of the test molecule during manufacture of lots comprising a therapeutic product comprising the test molecule, and wherein the computing device rejects any lot comprising the therapeutic product falling outside of the effective target range.

22. The non-transitory, tangible computer readable medium of claim 17, wherein: (a) the trial session is a clinical trial, the trial session dataset is a clinical trial dataset, and the set of subjects is a set of human clinical trial subjects as defined by the trial session dataset; (b) the trial session is a reactions based test, the trial session dataset is a reactions based dataset, and the set of subjects is a set of nonhuman mammals or cultured cells or tissues as defined by the reactions based dataset, optionally wherein the cultured cells comprise immune cells or members of a cell-based potency assay; or (c) the trial session is a trends based test, the trial session dataset is a trends based dataset, and the set of subjects is a set of subjects having attributes measured over time.

23. The non-transitory, tangible computer readable medium of claim 17, wherein a relative proximity estimation defines a difference in direction and / or distance from the origin for the first cluster and the second cluster, and wherein the difference correlates to a molecular attribute level of the test molecular attribute and a corresponding impact of the molecular attribute level on the biological response to the test molecule.

24. The non-transitory, tangible computer readable medium of claim 17, wherein the range of attribute exposure values corresponds to a dosing value or range of dosing values that results in the medically suitable biological response.

Images