MODELLING FRAMEWORK FOR EMBEDDING-BASED PREDICTIONS FOR COMPOUND-VIRAL PROTEIN ACTIVITY

§101 §103 §112

DETAILED ACTION Applicant’s response filed 07/20/2026 has been fully considered. Rejections and/or objections not reiterated from previous Office Actions are hereby withdrawn. The following rejections and/or objections are either reiterated or newly applied. Notice of Pre-AIA or AIA Status The present application, filed on or after March 16, 2013, is being examined under the first inventor to file provisions of the AIA . Status of the Claims Claims 1-9 are pending and under consideration in this action. Claims 10-15 are withdrawn from consideration pursuant to 37 CFR 1.142(b) as being drawn to a nonelected invention, there being no allowable generic or linking claim, in the amendment filed 04/08/2026. Priority The instant application claims domestic benefit to U.S. Provisional Application No. 63/193,845, filed 05/27/2021, as reflected in the filing receipt mailed 06/07/2022. The claim for domestic benefit for claims 1-9 is acknowledged. As such, the effective filing date of claims 1-9 is 05/27/2021. Specification Withdrawn Objections The objection to the abstract is withdrawn in view of Applicant’s amendments to the abstract filed 07/20/2026 (Applicant’s Remarks, Pg. 10). The objections to the Specification for minor informalities and use of trade names without accompanying generic terminology are withdrawn in view of Applicant’s amendments to the Specification filed 07/20/2026 (Applicant’s Remarks, Pg. 10). Newly Recited Objections The disclosure is objected to because it contains an embedded hyperlink and/or other form of browser-executable code (amended Para. [0013]; https://github.com/molecularsets/moses). Applicant is required to delete the embedded hyperlink and/or other form of browser-executable code; references to websites should be limited to the top-level domain name without any prefix such as http:// or other browser-executable code. See MPEP § 608.01. Claim Objections The objections to claims 1-2 and 5 are withdrawn in view of Applicant’s amendments to the claims filed 07/20/2026 (Applicant’s Remarks, Pg. 10). Claim Rejections - 35 USC § 112(a) The following is a quotation of the first paragraph of 35 U.S.C. 112(a): (a) IN GENERAL.—The specification shall contain a written description of the invention, and of the manner and process of making and using it, in such full, clear, concise, and exact terms as to enable any person skilled in the art to which it pertains, or with which it is most nearly connected, to make and use the same, and shall set forth the best mode contemplated by the inventor or joint inventor of carrying out the invention. The following is a quotation of the first paragraph of pre-AIA 35 U.S.C. 112: The specification shall contain a written description of the invention, and of the manner and process of making and using it, in such full, clear, concise, and exact terms as to enable any person skilled in the art to which it pertains, or with which it is most nearly connected, to make and use the same, and shall set forth the best mode contemplated by the inventor of carrying out his invention. Claims 8-9 are rejected under 35 U.S.C. 112(a) or 35 U.S.C. 112 (pre-AIA ), first paragraph, as failing to comply with the written description requirement. The claim(s) contains subject matter which was not described in the specification in such a way as to reasonably convey to one skilled in the relevant art that the inventor or a joint inventor, or for applications subject to pre-AIA 35 U.S.C. 112, the inventor(s), at the time the application was filed, had possession of the claimed invention. Any newly recited portion is necessitated by claim amendment. Claim 8 recites the limitation “administering a therapeutically effective dose of the certain compound to a human to treat a virus associated with the viral protein amino acid sequences examined”. The Specification (see Para. [0036]-[0037]) discloses that a certain compound is selected for administration or further clinical trials to treat the virus in a human. For example, Rifabutin may be selected for further trials or to treat SARS-COV-2 based on having a consistently low binding energy across the three proteins analyzed. A user administers a therapeutically effective dose of the certain compound selected to a human as part of a clinical trial of the compound in treating the virus, or to directly treat the virus in question. However, the claim is directed towards the administration of a therapeutically effective dose of any identified compound for any identified viral protein. The Specification does not provide any support for the physical administration of any identified compound to treat a virus associated with any examined protein sequence. Accordingly, the disclosure is not commensurate with the written description scope of the claim. Claim 9 is also rejected due to its dependency on claim 8. Response to Arguments under 35 U.S.C. 112(a) Applicant’s arguments filed 07/20/2026 have been fully considered but they are not persuasive. 1. Applicant argues that with regard to the recitation of "administering a therapeutically effective dose of the certain compound to a human to treat a virus associated with the viral protein amino acid sequences examined" in claim 8, the Office Action alleges that a skilled practitioner would be unable to directed to apply any identified compound to any identified viral protein. Applicant notes that the present disclosure is for "a modelling framework for embedding-based predictions for compound-viral protein activity" and "provides a new and innovative method for predicting activity value for compound-viral protein interactions" such that that "An aim of the provided method is to accurately identify a set of candidate compounds which can exhibit high activity against viral proteins and potentially inhibit them using novel in-silico techniques". Accordingly, the present disclosure does provide the skilled practitioner with any compound that is identified as being therapeutically effective for use with a specified viral protein, which may be from any virus of interest because the claimed model identifies the therapeutically effective interactions (Applicant’s Remarks, Pg. 11). It is respectfully submitted that this is not persuasive for the following reasons: Examiner notes that the arguments regarding determination of the therapeutically effective dose have been found persuasive. Accordingly, the written description rejection for claims 8 and 9 regarding the therapeutically effective dose have been withdrawn from the rejection above. MPEP § 2161.01(I) recites: The Federal Circuit has explained that a specification cannot always support expansive claim language and satisfy the requirements of 35 U.S.C. 112 "merely by clearly describing one embodiment of the thing claimed." LizardTech v. Earth Resource Mapping, Inc., 424 F.3d 1336, 1346, 76 USPQ2d 1731, 1733 (Fed. Cir. 2005). The issue is whether a person skilled in the art would understand the inventor to have invented, and been in possession of, the invention as broadly claimed. In LizardTech, claims to a generic method of making a seamless discrete wavelet transformation (DWT) were held invalid under 35 U.S.C. 112, first paragraph, because the specification taught only one particular method for making a seamless DWT and there was no evidence that the specification contemplated a more generic method. "[T]he description of one method for creating a seamless DWT does not entitle the inventor . . . to claim any and all means for achieving that objective." LizardTech, 424 F.3d at 1346, 76 USPQ2d at 1733. See also MPEP § 2163.05(I)(B): Tronzo v. Biomet, 156 F.3d at 1159, 47 USPQ2d at 1833 (Fed. Cir. 1998) (holding that the disclosure of a species in a parent application did not provide adequate written description support for claims to a genus in a child application where the specification taught against other species). Claim 8 still recites “administering a therapeutically effective dose of the certain compound to a human to treat a virus associated with the viral protein amino acid sequences examined”. As described in the rejection above, the Specification is silent on physical administration of any compound for any examined viral sequences. The Specification (see Para. [0037]) discloses that “a user administers a therapeutically effective dose of the certain compound selected per block to a human as part of a clinical trial of the compound in treating the virus, or to directly treat the virus in question. Data from the in vivo use of the compound to treat the virus may be collected and fed back into the data sets used by the consensus framework, and may be used a training data for the various ML models used by the consensus model to improve the efficacy and accuracy of the ML models in identifying different compounds for treating the virus in question in the future, or for identifying a compound to use in treating a different virus(including mutations or variants of the virus in question) in the future”. While block 360 indicates the administration of an effective dose of the compound, the Specification does not provide, for example, the physical administration to a patient nor does the Specification disclose an example of effective treatment of the specified virus for the identified compound (e.g., with Rifabutin for SARS-COV-2). Therefore, analogous to LizardTech and Tronzo cited above, the instant Specification provides support for a single embodiment/species (i.e., the administration of Rifabutin for SARS-COV-2) but does not provide support for the genus (i.e., the physical administration of any compound for any examined viral sequences). Similarly, the method of Anwar et al. (Combined Deep Learning and Molecular Docking Simulations Approach Identifies Potentially Effective FDA Approved Drugs for Repurposing Against SARS-CoV-2. ChemRxiv: 12227363.v1; published 5/4/2020; cited in the IDS dated 5/27/2022; previously cited), as described in the Claim Rejections – 35 U.S.C. 103 section below, describes a drug repurposing strategy specific to SARS-CoV-2, rather than a generic strategy for any viral disease (Anwar et al., see at least Abstract, and Pg. 25, Para. 2 – Pg. 26, Para. 1). Accordingly, the written description is not commensurate with the written description scope of claim 8, and this argument is not persuasive. Claim Rejections - 35 USC § 112(b) The rejection of claims 2-4 under 35 U.S.C. 112(b) is withdrawn in view of Applicant’s amendments to the claims filed 07/20/2026 (Applicant’s Remarks, Pg. 12). Claim Rejections - 35 USC § 101 35 U.S.C. 101 reads as follows: Whoever invents or discovers any new and useful process, machine, manufacture, or composition of matter, or any new and useful improvement thereof, may obtain a patent therefor, subject to the conditions and requirements of this title. Claims 1-9 are rejected under 35 U.S.C. 101 because the claimed invention is directed to an abstract idea without significantly more. The claims recite both (1) mathematical concepts (mathematical relationships, formulas or equations, or mathematical calculations) and (2) mental processes, i.e., concepts performed in the human mind (including observations, evaluations, judgements or opinions) (see MPEP § 2106.04(a)). Any newly recited portion is necessitated by claim amendment. Step 1: In the instant application, claims 1-9 are directed towards a method, which falls into one of the categories of statutory subject matter (Step 1: YES). Step 2A, Prong One: In accordance with MPEP § 2106, claims found to recite statutory subject matter (Step 1: YES) are then analyzed to determine if the claims recite any concepts that equate to an abstract idea, law of nature or natural phenomenon (Step 2A, Prong One). The following instant claims recite limitations that equate to one or more categories of judicial exceptions: Claim 1 recites a mathematical concept (i.e., nanomolar concentration values are calculated as the - log ⁡ ( a c t i v i t y ) , see Specification Para. [0016]) in “transforming bioactivities of the chemical data into nanomolar (nM) concentration values”; a mathematical concept (i.e., training an autoencoder involves optimizing parameters to minimize loss) in “training autoencoder models to generate numeric vector embedding representations for the SMILES representations and the viral protein amino acid sequences”; a mathematical concept in “providing the numeric vector embedding representations as features for downstream supervised learning of a plurality of regression models according to a plurality of different machine learning architectures”; a mental process (i.e., evaluating a compound-sequence pairing to determine an activity) in “generating a plurality of predicted activities for each pairing of one of the compounds and one of the viral protein amino acid sequences”; and a mental process (i.e., an evaluation of the top performing regression models) in “taking a consensus from a predefined number of top performing regression models selected from the plurality of regression models”. Claim 5 recites a mathematical concept in “wherein the plurality of different machine learning architectures include a Generalized Linear Model (GLM), Random Forests (RF), XGBoost, eXtreme Gradient Boosting (XGBoost), Support Vector Machines (SVM), Convolutional neural network (CNN), Long Short Term Memory (LSTM), CNN-LSTM, and Graph Attention Network (GAT)-CNN”. Claim 6 recites a mental process (i.e., an evaluation of the metrics to determine the best performing models) in “wherein the top performing regression models are determined from the plurality of regression models based on performance with respect evaluation metrics including at least one of: a mean absolute error; a mean squared error; a Pearson correlation R; and a coefficient of determination”. Claim 7 recites a mental process (i.e., an evaluation of the top performing regression models) in “wherein the predefined number of top performing regression models is five”. Claim 8 recites a mental process (i.e., an evaluation of binding energies to select a compound) in “selecting a certain compound from the list having a lowest binding energy for at least one of the viral protein amino acid sequences examined”. Claim 9 recites a mental process (i.e., an observation of the identified virus and compound) in “wherein the virus is SARS-COV-2 and the certain compound is Rifabutin”. These recitations are similar to the concepts of collecting information, and displaying certain results of the collection and analysis is Electric Power Group, LLC, v. Alstom (830 F.3d 1350, 119 USPQ2d 1739 (Fed. Cir. 2016)), comparing information regarding a sample or test to a control or target data in Univ. of Utah Research Found. v. Ambry Genetics Corp. (774 F.3d 755, 113 U.S.P.Q.2d 1241 (Fed. Cir. 2014)) and Association for Molecular Pathology v. USPTO (689 F.3d 1303, 103 U.S.P.Q.2d 1681 (Fed. Cir. 2012)), and organizing and manipulating information through mathematical correlations in Digitech Image Techs., LLC v Electronics for Imaging, Inc. (758 F.3d 1344, 111 U.S.P.Q.2d 1717 (Fed. Cir. 2014)) that the courts have identified as concepts that can be practically performed in the human mind or mathematical relationships. The abstract ideas recited in the claims are evaluated under the broadest reasonable interpretation (BRI) of the claim limitations when read in light of and consistent with the specification, and are determined to be directed to mental processes that in the simplest embodiments are not too complex to practically perform in the human mind. Additionally, the recited limitations that are identified as judicial exceptions from the mathematical concepts grouping of abstract ideas are abstract ideas irrespective of whether or not the limitations are practical to perform in the human mind. Specifically, claim 1 involves nothing more than determining concentration values, training autoencoder models to generate numeric vector embedding representations, using embedding representations in regression models, generating predicted activities, and taking a consensus for the top performing models. The steps reciting determining concentration values, training autoencoder models to generate numeric vector embedding representations, and using embedding representations in regression models are, under the BRI, performed using mathematical operations. The instant Specification (see Para. [0016]) discloses that the values presented in PubChem standards are converted to a standardized pChEMBL value where p C h E M B L   =   - log 10 ⁡ ( A c t i v i t y P u b C h e m )   +   6 , where ActivityPubChem corresponds to one of IC50, EC50, AC50, Ki, Kd, or Potency. The instant Specification (see Para. [0014] and [0016]) also discloses that molecule datasets are used to generate a TF-LSTM based autoencoder and viral protein sequences are used to train a CNN based autoencoder. The instant Specification (see Para. [0032]) also discloses that the plurality of regression models includes a Generalized Linear Model (GLM), Random Forests (RF), XGBoost, Support Vector Machines (SVM), Convolutional neural network (CNN), Long Short Term Memory (LSTM), CNN-LSTM, and Graph Attention Network (GAT)-CNN. Additionally, since there are no specifics in the methodology, the steps of generating predicted activities and taking a consensus of the top performing models, are something that under BRI, one could perform mentally. Therefore, the claimed steps are not further defined beyond something that reads on performing a calculation using a computer as a tool, and merely looking at data and making a determination. As such, said steps are directed to judicial exceptions. The instant claims must therefore be examined further to determine whether they integrate the abstract idea into a practical application (Step 2A, Prong One: YES). Step 2A, Prong Two: In determining whether a claim is directed to a judicial exception, further examination is performed that analyzes if the claim recites additional elements that when examined as a whole integrates the judicial exception(s) into a practical application (MPEP § 2106.04(d)). A claim that integrates a judicial exception into a practical application will apply, rely on, or use the judicial exception in a manner that imposes a meaningful limit on the judicial exception. The claimed additional elements are analyzed to determine if the abstract idea is integrated into a practical application (MPEP § 2106.04(d)(I)). If the claim contains no additional elements beyond the abstract idea, the claim fails to integrate the abstract idea into a practical application (MPEP § 2106.04(d)(III)). The following independent claims recite limitations that equate to additional elements: Claim 1 recites “collecting chemical data including: simplified molecular-input line-entry system (SMILES) representations of compounds; viral protein amino acid sequences; and compound-viral protein activities”; and “outputting a list of the compounds according to the consensus having a highest predicted activity against the viral protein amino acid sequences”. Regarding the above cited limitation in claim 1 of (i) collecting chemical data including: simplified molecular-input line-entry system (SMILES) representations of compounds; viral protein amino acid sequences; and compound-viral protein activities. This limitation equates to insignificant, extra-solution activity of mere data gathering because this limitation gathers data before the recited judicial exceptions of determining concentration values, training autoencoder models to generate numeric vector embedding representations, using embedding representations in regression models, generating predicted activities, and taking a consensus for the top performing models (see MPEP § 2106.04(d)). Regarding the above cited limitation in claim 1 of (ii) outputting a list of the compounds according to the consensus having a highest predicted activity against the viral protein amino acid sequences. This limitation equates to an extra-solution step of generally outputting a result, which is incidental to the primary process of determining concentration values, training autoencoder models to generate numeric vector embedding representations, using embedding representations in regression models, generating predicted activities, and taking a consensus for the top performing models (see MPEP § 2106.05(g)). Additionally, none of the recited dependent claims recite additional elements which would integrate the judicial exception into a practical application. Specifically, claims 2-4 further limit the collection of chemical data; and claim 8 recites a generic administration step that equates to an extra-solution “apply it” step (see MPEP § 2106.05(f)). As such, claims 1-9 are directed to an abstract idea (Step 2A, Prong Two: NO). Step 2B: Claims found to be directed to a judicial exception are then further evaluated to determine if the claims recite an inventive concept that provides significantly more than the judicial exception itself (Step 2B). The claims do not include additional elements that are sufficient to amount to significantly more than the judicial exception. The instant independent claims recite the same additional elements described in Step 2A, Prong Two above. Regarding the above cited limitation in claim 1 of (i) collecting chemical data including: simplified molecular-input line-entry system (SMILES) representations of compounds; viral protein amino acid sequences; and compound-viral protein activities. This limitation does not include any specific steps for collecting the chemical data. Under the BRI, this limitation is merely receiving data for the subsequent steps of determining concentration values, training autoencoder models to generate numeric vector embedding representations, using embedding representations in regression models, generating predicted activities, and taking a consensus for the top performing models. Therefore, these limitations equate to receiving/transmitting data over a network, which the courts have established as a WURC limitation of a generic computer in buySAFE, Inc. v. Google, Inc., 765 F.3d 1350, 1355, 112 USPQ2d 1093, 1096 (Fed. Cir. 2014). Regarding the above cited limitation in claim 1 of (ii) outputting a list of the compounds according to the consensus having a highest predicted activity against the viral protein amino acid sequences. This limitation when viewed individually and in combination, is a well-understood, routine, and conventional (WURC) limitation as taught by Anwar et al. (Combined Deep Learning and Molecular Docking Simulations Approach Identifies Potentially Effective FDA Approved Drugs for Repurposing Against SARS-CoV-2. ChemRxiv: 12227363.v1. (51 pages); cited in the IDS dated 5/27/2022; previously cited). Anwar et al. discloses an approach combining deep learning and molecular docking to identify the most promising FDA approved candidates that can be repurposed to treat COVID-19. The output is a list of 49 most promising candidates with the best consensus KIBA scores and docking binding affinity values (limitation (ii)) (Abstract). These additional elements do not comprise an inventive concept when considered individually or as an ordered combination that transforms the claimed judicial exception into a patent-eligible application of the judicial exception. Therefore, the instant claims do not amount to significantly more than the judicial exception itself (Step 2B: NO). As such, claims 1-9 are not patent eligible. Response to Arguments under 35 U.S.C. 101 Applicant’s arguments filed 07/20/2026 have been fully considered but they are not persuasive. 1. Applicant argues that the Specification describes “Improvements to computer component or system performance based upon adjustments of parameters of a machine learning model associated with tasks or workstreams” and offers “an improved way of training the machine learning model”. See Specification at [0018]-[0019] as well as [0008] and [0012]) These features are set forth in claim 1, which recites, in part, "training autoencoder models to generate numeric vector embedding representations for the SMILES representations and the viral protein amino acid sequences; providing the numeric vector embedding representations as features for downstream supervised learning of a plurality of regression models according to a plurality of different machine learning architectures; generating a plurality of predicted activities for each pairing of one of the compounds and one of the viral protein amino acid sequences; taking a consensus from a predefined number of top performing regression models selected from the plurality of regression models". Accordingly, claim 1 (and the dependent claims thereof) recite substantially more than an abstract idea, and provide for impartments to computer-based technologies, which are an integration of the concepts into a practical application (Applicant’s Remarks, Pg. 12-14). It is respectfully submitted that this is not persuasive for the following reasons: In Ex Parte Desjardins, at “Step 2A Prong Two, the ARP then determined that the specification identified improvements as to how the machine learning model itself operates, including training a machine learning model to learn new tasks while protecting knowledge about previous tasks to overcome the problem of “catastrophic forgetting” encountered in continual learning systems. Importantly, the ARP evaluated the claims as a whole in discerning at least the limitation “adjust the first values of the plurality of parameters to optimize performance of the machine learning model on the second machine learning task while protecting performance of the machine learning model on the first machine learning task” reflected the improvement disclosed in the specification. Accordingly, the claims as a whole integrated what would otherwise be a judicial exception instead into a practical application at Step 2A Prong Two.” Therefore, in Ex Parte Desjardins, the specific training limitation of “adjusting the first values…” reflects the improvement to overcome the problem of “catastrophic forgetting”. Ex Parte Desjardins is different from the instant case for the following reasons: The instant Specification (see Para. [0019]) discloses that the “models may use different inputs, and reach conclusions in different ways, and the consensus between the models ensures accuracy in the prediction generated under the consensus framework”. With regards to training, the instant Specification (see Para. [0014] and [0016]) discloses that the SMILES compound set is used to train a TF-LSTM based autoencoder and that the viral protein sequence set is used to train CNN based autoencoder. The Specification therefore discloses that trained representations are generated as input for the various regression models evaluated in the consensus framework, reflected in the limitations of “training autoencoder models to generate numeric vector embedding representations…”, and “providing the numeric vector embedding representations as features for downstream supervised learning of a plurality of regression models…”. However, other than merely disclosing that the consensus model is cost-effective and time-efficient (see Specification, Para. [0008] and [0012]), the Specification does not appear to provide any indication (e.g., steps or examples) of a specific training process for the autoencoders for SMILES strings and viral protein sequences to improve the downstream regression models, the computer components, or the system performance. Accordingly, the cited limitations do not reflect an improvement in the functioning of a computer, or an improvement to other technology or a technical field. Additionally, the limitations of, “generating a plurality of predicted activities…”, and “taking a consensus from a predefined number of top performing regression models…”, indicated by Applicant, also do not reflect an improvement in the training of the machine learning model nor an improvement in the functioning of a computer. This argument is thus not persuasive. 2. Applicant argues that regarding claims 8-9, Applicant respectfully notes that the particular treatment or prophylaxis of a disease or medical condition has been found by the courts to be indicative that an additional element (or combination of elements) to have integrated the exception into a practical application, that renders the claim as a whole directed to patent eligible subject matter. See, MPEP §§ 2106.04(d)(2) and 2106.05(e). (Applicant’s Remarks, Pg. 14). It is respectfully submitted that this is not persuasive for the following reasons: MPEP § 2106.04(d)(2) recites: The treatment or prophylaxis limitation must be "particular," i.e., specifically identified so that it does not encompass all applications of the judicial exception(s). For example, consider a claim that recites mentally analyzing information to identify if a patient has a genotype associated with poor metabolism of beta blocker medications. This falls within the mental process grouping of abstract ideas enumerated in MPEP § 2106.04(a). The claim also recites "administering a lower than normal dosage of a beta blocker medication to a patient identified as having the poor metabolizer genotype." This administration step is particular, and it integrates the mental analysis step into a practical application. Conversely, consider a claim that recites the same abstract idea and "administering a suitable medication to a patient." This administration step is not particular, and is instead merely instructions to "apply" the exception in a generic way. Thus, the administration step does not integrate the mental analysis step into a practical application. With regards to claim 8, the limitation of “administering a therapeutically effective dose of the certain compound to a human to treat a virus associated with the viral protein amino acid sequences examined” is not particular because there is no limitation on the compound or the targeted virus, as claim 1 recites the collection of both SMILES representations of compounds and viral protein amino acid sequences (and thereby the administration is for any of those compounds for any of the recited viral sequences). Therefore, the administration is not particular and recites instructions to “apply” the exception in a generic way. With regards to claim 9, “wherein the virus is SARS-COV-2 and the certain compound is Rifabutin”. This limitation recites a particular administration of a therapeutically effective dose of Rifabutin for the treatment of SARS-COV-2. However, claim 9 is dependent on claim 8 (and claim 1), which both contain limitations reciting judicial exceptions. Accordingly, claims 8 and 9 are not patent-eligible and this argument is not persuasive. Claim Rejections - 35 USC § 103 In the event the determination of the status of the application as subject to AIA 35 U.S.C. 102 and 103 (or as subject to pre-AIA 35 U.S.C. 102 and 103) is incorrect, any correction of the statutory basis (i.e., changing from AIA to pre-AIA ) for the rejection will not be considered a new ground of rejection if the prior art relied upon, and the rationale supporting the rejection, would be the same under either status. The following is a quotation of 35 U.S.C. 103 which forms the basis for all obviousness rejections set forth in this Office action: A patent for a claimed invention may not be obtained, notwithstanding that the claimed invention is not identically disclosed as set forth in section 102, if the differences between the claimed invention and the prior art are such that the claimed invention as a whole would have been obvious before the effective filing date of the claimed invention to a person having ordinary skill in the art to which the claimed invention pertains. Patentability shall not be negated by the manner in which the invention was made. The factual inquiries for establishing a background for determining obviousness under 35 U.S.C. 103 are summarized as follows: 1. Determining the scope and contents of the prior art. 2. Ascertaining the differences between the prior art and the claims at issue. 3. Resolving the level of ordinary skill in the pertinent art. 4. Considering objective evidence present in the application indicating obviousness or nonobviousness. This application currently names joint inventors. In considering patentability of the claims the examiner presumes that the subject matter of the various claims was commonly owned as of the effective filing date of the claimed invention(s) absent any evidence to the contrary. Applicant is advised of the obligation under 37 CFR 1.56 to point out the inventor and effective filing dates of each claim that was not commonly owned as of the effective filing date of the later invention in order for the examiner to consider the applicability of 35 U.S.C. 102(b)(2)(C) for any potential 35 U.S.C. 102(a)(2) prior art against the later invention. 1. Claims 1-4, 6, and 8-9 are rejected under 35 U.S.C. 103 as being unpatentable over Anwar et al. (Combined Deep Learning and Molecular Docking Simulations Approach Identifies Potentially Effective FDA Approved Drugs for Repurposing Against SARS-CoV-2. ChemRxiv: 12227363.v1; published 5/4/2020; cited in the IDS dated 5/27/2022; previously cited) in view of Zakharov et al. (Novel Consensus Architecture to Improve Performance of Large-Scale Multitask Deep Learning QSAR Models. J. Chem. Inf. Model. 59(11): 4613-4624 (2019); published 10/4/2019; previously cited), Gómez-Bombarelli et al. (Automatic Chemical Design using a Data-Driven Continuous Representation of Molecules. ACS Cent. Sci. 4: 268-276 (2018); published 01/12/2018; newly recited), and Li et al. (De novo Prediction of Moonlighting Proteins Using Multimodal Deep Ensemble Learning. Front Genet. 12: 630379 (12 pages) (2021); published 03/21/2021; newly recited). This rejection is newly recited and necessitated by claim amendment. Regarding claim 1, Anwar et al. teaches a method for combining deep learning with molecular docking experiments to identify the most promising candidates from a list of FDA approved drugs that can be repurposed to treat COVID-19 (Abstract). Anwar et al. further teaches that their deep learning model automatically incorporates useful and required features from raw ligands and proteins into the model to predict the drug and protein interactions. They have utilized 1-D sequences of ligands (SMILES) and proteins (FASTA) to train the model. SMILES sequences were retrieved from the Canadian Drug Bank and FASTA sequences of 24 viral proteins were acquired from the published genome of SARS-CoV-2 available and C-I-TASSER (i.e., collecting chemical data including simplified molecular-input line-entry system (SMILES) representations of compounds and viral protein amino-acid sequences) (Pg. 8, Para. 4 – Pg. 9, Para. 1, Pg. 11, Fig. 1, and Pg. 12, Para. 1). Anwar et al. further teaches that they used a benchmark KIBA dataset for training the model and predicting binding affinities. The KIBA dataset comprises selectively assays of the kinase proteins family and the associated inhibitors. The KIBA values have been contrived by combining Ki, Kd, and IC50 values from several sources (i.e., collecting chemical data including compound-viral protein activities) (Pg. 9, Para. 2). Anwar et al. further teaches that after training the model, they predicted the KIBA scores of FDA approved and investigational drugs. The drugs were screened against 24 viral proteins from the published genome of SARS-CoV-2 (i.e., generating a plurality of predicted activities for each pairing of one of the compounds and one of the viral protein amino acid sequences) (Pg. 12, Para. 1). Anwar et al. further teaches that FDA approved drugs were ranked according to their KIBA scores for each viral protein (Pg. 12, Para. 2). They also performed docking simulations for selected drugs with high predicted KIBA scores against the SARS-CoV-2 viral proteins to determine binding affinity values. Docking binding affinity values were plotted against KIBA scores for the drugs. Drugs with the best consensus KIBA scores and binding affinity values were selected, giving an output list of the top 49 FDA approved drugs that are predicted to effectively inhibit selected SARS-CoV-2 viral proteins (i.e., outputting a list of the compounds according to the consensus having a highest predicted activity against the viral protein amino acid sequences) (Pg. 16, Para. 3). Regarding claim 2, Anwar et al. teaches that SMILES sequences for both FDA approved and investigational drugs were retrieved from DrugBank (i.e., wherein the SMILES representations of compounds are collected from publicly available databases) (Pg. 12, Para. 1). Regarding claim 3, Anwar et al. teaches that the FASTA sequences of 24 viral proteins have been acquired from the published genome of SARS-CoV-2 available at C-I-TASSER (i.e., wherein the viral protein amino acid sequences are collected from a publicly available database) (Pg. 12, Para. 1). Regarding claim 4, Anwar et al. teaches that the KIBA dataset comprises selectivity assays of the kinase proteins family and the associated inhibitors. It predominantly embodies the corresponding KIBA scores. The KIBA values have been contrived by combining Ki, Kd, and IC50 values (i.e., wherein the compound-viral protein activities are collected from publicly available databases) (Pg. 9, Para. 2). Regarding claim 8, Anwar et al. teaches that drugs with the best consensus KIBA scores and binding affinities were selected (Pg. 16, Para. 3).The compounds in Table 1 have the best KIBA scores and lowest docking binding affinity values against selected SARS-CoV-2 viral proteins (i.e., selecting a certain compound from the list having a lowest binding energy for at least one of the viral protein amino acid sequences examined) (Pg. 38-45, Table 1). Anwar et al. further teaches that given the potential to inhibit 3CLpro, ease of oral administration and anticoagulant activity, edoxaban appears to be a promising candidate drug for treating COVID-19 (Pg. 24, Para. 2). Edoxaban is one of the identified compounds with the best consensus KIBA score and binding affinities (i.e., administering a therapeutically effective dose of the certain compound to a human to treat a virus associated with the viral protein amino acid sequences examined) (see Pg. 44, Table 1). Regarding claim 9, Anwar et al. teaches that Rifabutin ranked among the FDA approved drugs with the best predicted values against SARS-CoV-2 Viral Proteins (Pg. 38, Table 1). With the potential of inhibiting RdRp and helicase amongst other SARS-CoV-2 viral proteins, rifabutin can be a promising therapeutic option (i.e., wherein the virus is SARS-COV-2, and the certain compound is rifabutin) (Pg. 23, Para. 2). Anwar et al. does not teach transforming bioactivities of the chemical data into nanomolar (nM) concentration values (claim 1); training autoencoder models to generate numeric vector embedding representations for the SMILES representations and the viral protein amino acid sequences (claim 1); providing the numeric vector embedding representations as features for downstream supervised learning of a plurality of regression models according to a plurality of different machine learning architectures (claim 1); taking a consensus from a predefined number of top performing regression models selected from the plurality of regression models (claim 1); and wherein the top performing regression models are determined from the plurality of regression models based on performance with respect evaluation metrics including at least one of: a mean absolute error, a mean squared error, a Pearson correlation R; and a coefficient of determination (claim 6). Regarding claim 1, Zakharov et al. teaches a deep learning consensus architecture that combines consensus and multitask deep learning approaches together to generate large-scale QSAR models (Abstract). Zakharov et al. further teaches that bioactivities were extracted from the ChEMBL database. All IC50 values were converted into p I C 50   =   - l o g   ( I C 50 , [ M ] ) . Compounds were kept with the following conditions: (i) assay confidence score more than 6, (ii) operator of activity equals to “=”, (iii) bioactivity expressed in “IC50” values, and (iv) activity unit is “nM” (i.e., transforming bioactivities of the chemical data into nanomolar (nM) concentration values) (Pg. 4615, Col. 1, Para. 2). Zakharov et al. further teaches the construction of a consensus models for random forest and deep learning models. Four consensus models were constructed: (i) consensus Random Forest model (Consensus_RF), (ii) consensus multitask deep learning model (Consensus_MDL), (iii) consensus Random Forest for proteochemometrics (Consensus_PCM_RF), and (iv) consensus deep learning model for proteochemometrics (Consensus_PCM_DL) (i.e., taking a consensus from a predefined number of top performing regression models selected from the plurality of regression models) (Pg. 4616, Col. 1, Para. 4). Regarding claim 1, Gómez-Bombarelli et al. teaches a method to convert discrete representations of molecules to and from a multidimensional continuous representation (Abstract). Gómez-Bombarelli et al. further teaches the structure of the autoencoder, which uses a SMILES string as input. The encoder converts each molecule to a vector in latent space (Pg. 269, Fig. 1). Gomez-Bombarelli et al. further teaches the training of the autoencoder (i.e., training autoencoder models to generate numeric vector embedding representations for the SMILES representations) (Pg. 274, Col. 2, Para. 4-6). Regarding claim 1, Li et al. teaches a workflow to extract features from protein sequence FASTA input files. A deep auto-encoder is applied for secondary structure features (Pg. 5, Fig. 1). Li et al. further teaches that the output is a feature vector (Pg. 6, Col. 2, Para. 3). Li et al. further teaches the training of the auto-encoder (i.e., training autoencoder models to generate numeric vector embedding representations for the viral protein amino acid sequences) (Pg. 6, Col. 2, Para. 6 – Pg. 7, Col. 1, Para. 1). Li et al. further teaches that the extracted feature vectors are integrated and used in downstream machine learning classifiers including logistic regression, random forest, and support vector machine (i.e., providing the numeric vector embedding representations as features for downstream supervised learning of a plurality of regression models according to a plurality of different machine learning architectures) (Pg. 5, Fig. 1; and Pg. 5, Col. 1, Para. 1 – Pg. 6, Col. 1, Para. 3). Regarding claim 6, Zakharov et al. teaches that the regression models were evaluated using squared Pearson’s coefficient and root mean square error (i.e., wherein the top performing regression models are determined from the plurality of regression models based on performance with respect evaluation metrics including at least one of: a Pearson correlation R) (Pg. 4616, Col. 2, Para. 2 – Pg. 4616, Col. 1, Para. 1). Therefore, regarding claims 1-4, 6, and 8-9, it would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to modify the deep learning method evaluating compounds for drug repurposing of Anwar et al. with the consensus model of Zakharov et al. because consensus models have been shown to have reduced variability compared to using individual models, therefore consensus models provide more reliable and accurate predictions (Zakharov et al., Pg. 4614, Col. 1, Para. 3). One of ordinary skill in the art would be able to combine the teachings of Anwar et al. with Zakharov et al. with reasonable expectation of success due to the same nature of the problem to be solved, since both incorporate a method for using machine learning to predict compound activity. It would have also been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to modify the deep learning method evaluating compounds for drug repurposes of Anwar et al. with the autoencoder for SMILES strings of Gómez-Bombarelli et al. because the autoencoder of Gómez-Bombarelli et al. captures characteristic features of the molecular training set and provides good predictive power when training jointly with a property prediction task (e.g., as in instant claim 1) (Gómez-Bombarelli et al., Pg. 274, Col. 1, Para. 3). One of ordinary skill in the art would be able to combine the teachings of Anwar et al. with Gómez-Bombarelli et al. with reasonable expectation of success due to the same nature of the problem to be solved, since both incorporate a method for generating a representation of a SMILES string. It would have also been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to modify the deep learning method evaluating compounds for drug repurposes of Anwar et al. with the autoencoder for FASTA sequences of Li et al. because the use of an autoencoder for extracting features improves the performance of the downstream classifiers (Li et al., Pg. 6, Col. 1, Para. 3; and Pg. 6,Table 1). One of ordinary skill in the art would be able to combine the teachings of Anwar et al. with Li et al. with reasonable expectation of success due to the same nature of the problem to be solved, since both incorporate a method for generating representations using FASTA sequences. Therefore, regarding claims 1-4, 6, and 8-9, the instant invention is prima facie obvious (MPEP § 2142). 2. Claim 5 is rejected under 35 U.S.C. 103 as being unpatentable over Anwar et al. in view of Zakharov et al., Gómez-Bombarelli et al., and Li et al. as applied to claims 1-4, 6, and 8-9 above, and further in view of Blanco et al. (Prediction of high anti-angiogenic activity peptides in silico using a generalized linear model and feature selection. Sci Rep. 8: 15688 (11 pages) (2018); published 10/24/2018; previously cited) and Lin et al. (DeepGS: Deep Representation Learning of Graphs and Sequences for Drug-Target Binding Affinity Prediction. ECAI 2020, 24th European Conference on Artificial Intelligence. Pg. 1301-1308 (2020); published 9/18/2020; previously cited). This rejection is newly recited and necessitated by claim amendment. Regarding claim 5, Anwar et al. teaches that the DeepDTA model usings Bi-LSTM and CNN blocks which combine to eventually predict the drug-protein binding affinities (i.e., wherein the plurality of different machine learning architectures includes a Convolutional neural network (CNN), Long Short Term Memory (LSTM), and CNN-LSTM) (Pg. 11, Fig. 1 (caption on Pg. 12)). Regarding claim 5, Zakharov et al. teaches the use of a random forest model for activity evaluation (i.e., wherein the plurality of different machine learning architectures include a random forest (RF)) (Pg. 4615, Col. 2, Para. 4-5). Anwar et al. in view of Zakharov et al., Gómez-Bombarelli et al., and Li et al., as applied to claims 1-4, 6, and 8-9 above, does not teach wherein the plurality of different machine learning architectures include a Generalized Linear Model (GLM), eXtreme Gradient Boosting (XGBoost), Support Vector Machines (SVM), and Graph Attention Network (GAT)-CNN. Regarding claim 5, Lin et al. teaches a model drug-target binding affinity prediction that consists of three major building blocks. One of the major blocks learns low-dimension vector representations for target/protein sequences, using a convolutional neural network (CNN). The other two blocks learn two representations for drugs, by using a graph attention network (GAT) and a bi-directional gate recurrent unit (BiGRU), respectively. Specifically, (i) the CNN and BiGRU blocks extract local chemical context information of amino acids in targets and atoms in drugs, respectively. Since the label/one-hot encodings of amino acids and atoms often neglect the context information, and motivated by the idea of Word2Vec, they leverage advanced techniques, Smi2Vec and Prot2Vec, to encode the amino acids and atoms to a distributed representation, before plugging them to CNN and BiGRU. (ii) The newly designed GAT-based molecular structure modeling approach extracts the topological features of drugs, by aggregating the representations of r-radius sub-graphs. (iii) The learned representations for both drugs and targets are then passed to a neural network to predict the binding affinity (i.e., wherein the plurality of different machine learning architectures include a Graph Attention Network (GAT)-CNN) (Title, Pg. 1301, Col. 2, Para. 3). Lin et al. further teaches comparison of the model to SimBoost, which trains a gradient boosting machine model to represent the nonlinear associations between the input features and the binding affinities (i.e., wherein the plurality of different machine learning architectures include eXtreme Gradient Boosting (XGBoost) (Pg. 1304, Col. 2, Para. 1). Regarding claim 5, Blanco et al. teaches a method to predict peptide activity using machine learning (Abstract). Blanco et al. further teaches that following machine learning algorithms were evaluated: random forest (RF), K-nearest neighbors (k-NN), support vector machine (SVM), and a generalized linear model (glmnet) (i.e., wherein the plurality of different machine learning architectures include a generalized linear model (GLM), random forest (RF), and support vector machine (SVM)) (Pg. 8, Para. 4). Therefore, regarding claim 5, it would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to modify the deep learning method evaluating compounds for drug repurposing of Anwar et al. in view of Zakharov et al., Gómez-Bombarelli et al., and Li et al. with the evaluation of additional machine learning architectures as described by Lin et al. and Blanco et al. because the method of Blanco et al. extensively evaluates the various machine learning architectures using nested resampling, cross-validation, hyperparameter tuning and different runs to determine the top performing model (Blanco et al., Abstract). Additionally, adding an attention mechanism to concatenate drug and protein representations for prediction (i.e., the method of Lin et al.) improves the binding affinity prediction by considering both local chemical context and topological information of drugs at the same time (Lin et al, Pg. 1307, Col. 2, Para. 2). One of ordinary skill in the art would be able to combine the teachings of Anwar et al. in view of Zakharov et al., Gómez-Bombarelli et al., and Li et al. with Lin et al. and Blanco et al. with reasonable expectation of success due to the same nature of the problem to be solved, since all three incorporate a method for predicting activity using machine learning. Therefore, regarding claim 5, the instant invention is prima facie obvious (MPEP § 2142). 3. Claim 7 is rejected under 35 U.S.C. 103 as being unpatentable over Anwar et al. in view of Zakharov et al., Gómez-Bombarelli et al., and Li et al. as applied to claims 1-4, 6, and 8-9 above, and further in view of Verras et al. (Shared Consensus Machine Learning Models for Predicting Blood Stage Malaria Inhibition. J. Chem. Inf. Model. 57(3): 445-453 (2017); published 3/3/2017; previously cited). This rejection is newly recited and necessitated by claim amendment. Anwar et al. in view of Zakharov et al., Gómez-Bombarelli et al., and Li et al., as applied to claims 1-4, 6, and 8-9 above, does not teach wherein the predefined number of top performing regression models is five. Regarding claim 7, Verras et al. teaches a method for developing a consensus model that has increased predictive power over individual methods to screen compound libraries for potential inhibitory activity (Abstract). Verras et al. further teaches that to evaluate whether consensus models can perform better than individual models, a simple averaging was done of the Bayesian prediction from each model. They generated 255 combinations of leave-x-models-out, where x varies from 0 to 7, and took the mean for all predictions in each set (Pg. 447, Col. 2, Para. 4). The best performing consensus model in validation against the St. Jude Vendor Library set was MMV Model 1, MMV Model 4, AZ Model 1, GSK Model 1, and Nov Model 1 (ROC AUC 0.854) (i.e., wherein the predefined number of top performing regression models is five) (Pg. 449, Col. 2, Para. 1). Therefore, regarding claim 7, it would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to modify the deep learning method evaluating compounds for drug repurposing of Anwar et al. in view of Zakharov et al., Gómez-Bombarelli et al., and Li et al. with the consensus analysis of Verras et al. because the development of consensus models yields increased predictive power over any single model (Verras et al., Abstract). One of ordinary skill in the art would be able to combine the teachings of Anwar et al. in view of Zakharov et al., Gómez-Bombarelli et al., and Li et al. with Verras et al. with reasonable expectation of success due to the same nature of the problem to be solved, since both incorporate a consensus analysis for model evaluation. Therefore, regarding claim 7, the instant invention is prima facie obvious (MPEP § 2142). Response to Arguments under 35 U.S.C. 103 Applicant’s arguments filed 07/20/2026 have been fully considered but they are not persuasive. 1. Applicant argues that none of the cited references, alone or in combination provide the features of “training autoencoder models to generate numeric vector embedding representations for the SMILES representations and the viral protein amino acid sequences; providing the numeric vector embedding representations as features for downstream supervised learning of a plurality of regression models according to a plurality of different machine learning architectures”, or in the alternative, no prima facie case has been presented for how such features could be derived from the references. (Applicant’s Remarks, Pg. 14-15). It is respectfully submitted that this is not persuasive for the following reasons: Applicant’s arguments regarding the obviousness of the claims over the cited references as failing to teach the claimed training of autoencoder models and subsequent use of the vector embedding representations as features for downstream supervised learning, as in amended claim 1 have been considered but they are not persuasive in view of the new grounds of rejection that relies on a new combination of references as necessitated by claim amendment. Conclusion No claims allowed. Applicant's amendment necessitated the new ground(s) of rejection presented in this Office action. Accordingly, THIS ACTION IS MADE FINAL. See MPEP § 706.07(a). Applicant is reminded of the extension of time policy as set forth in 37 CFR 1.136(a). A shortened statutory period for reply to this final action is set to expire THREE MONTHS from the mailing date of this action. In the event a first reply is filed within TWO MONTHS of the mailing date of this final action and the advisory action is not mailed until after the end of the THREE-MONTH shortened statutory period, then the shortened statutory period will expire on the date the advisory action is mailed, and any nonprovisional extension fee (37 CFR 1.17(a)) pursuant to 37 CFR 1.136(a) will be calculated from the mailing date of the advisory action. In no event, however, will the statutory period for reply expire later than SIX MONTHS from the mailing date of this final action. Inquiries Any inquiry concerning this communication or earlier communications from the examiner should be directed to DIANA P SANFORD whose telephone number is (571)272-6504. The examiner can normally be reached Mon-Fri 8am-5pm EST. Examiner interviews are available via telephone, in-person, and video conferencing using a USPTO supplied web-based collaboration tool. To schedule an interview, applicant is encouraged to use the USPTO Automated Interview Request (AIR) at http://www.uspto.gov/interviewpractice. If attempts to reach the examiner by telephone are unsuccessful, the examiner’s supervisor, Karlheinz Skowronek can be reached at (571)272-9047. The fax phone number for the organization where this application or proceeding is assigned is 571-273-8300. Information regarding the status of published or unpublished applications may be obtained from Patent Center. Unpublished application information in Patent Center is available to registered users. To file and manage patent submissions in Patent Center, visit: https://patentcenter.uspto.gov. Visit https://www.uspto.gov/patents/apply/patent-center for more information about Patent Center and https://www.uspto.gov/patents/docx for information about filing in DOCX format. For additional questions, contact the Electronic Business Center (EBC) at 866-217-9197 (toll-free). If you would like assistance from a USPTO Customer Service Representative, call 800-786-9199 (IN USA OR CANADA) or 571-272-1000. /D.P.S./Examiner, Art Unit 1687 /Lori A. Clow/Primary Examiner, Art Unit 1687