DETAILED ACTION
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 .
Claim Status
Claim 13 is cancelled.
Claim 14 is newly added.
Claims 1-12 and 14 are currently pending and under exam herein.
Claims 1-12 and 14 are rejected.
Claim 8 is objected to.
Priority
The instant application is the U.S. National Stage of International Application No. PCT/EP2021/070646 filed 23 July 2021, which claims priority to European application No. 20305852.4 filed 24 July 2020. The claimed priority is acknowledged. However, the prior applications do not contain adequate support for claim 14 because the applications do not disclose a “computer-readable support.” At this point in examination, the effective filing date of claims 1-12 is 24 July 2020 and the effective filing date of claim 14 is 19 January 2023.
Information Disclosure Statement
The information disclosure statement filed 8 January 2025 fails to comply with 37 CFR 1.98(a)(3)(i) because it does not include a concise explanation of the relevance, as it is presently understood by the individual designated in 37 CFR 1.56(c) most knowledgeable about the content of the information, of each reference listed that is not in the English language. It has been placed in the application file, but the information referred to therein has not been considered.
The information disclosure statements (IDS) submitted on 19 January 2023 and 4 April 2025 comply with 37 CFR 1.98. Accordingly, all references listed have been considered by the examiner.
Drawings
The drawings filed on 19 January 2023 have been received and are accepted.
Claim Objections
Claim 8 is objected to because it recites “second classifier formed a Support Vector Machine,” which should read “second classifier formed by a Support Vector Machine.” Appropriate correction is required.
Claim Rejections - 35 USC § 112
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.
Claim 14 is 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 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. The term “non-transitory computer-readable support” does not appear anywhere in the specification. The specification only refers to a “computing device” comprising “a computer” and “a memory” at para. [0055]. There is no disclosure that describes what constitutes a “support.” The claim encompasses any non-transitory medium capable of storing code instructions, which potentially includes paper printouts, magnetic tapes, optical discs, solid-state memory, or other unspecified media, with the specification providing no description showing possession of this breadth. Therefore, the specification fails to demonstrate that the inventor was in possession of the full scope of the claimed “non-transitory computer-readable support” at the time of filing.
The following is a quotation of 35 U.S.C. 112(b):
(b) CONCLUSION.—The specification shall conclude with one or more claims particularly pointing out and distinctly claiming the subject matter which the inventor or a joint inventor regards as the invention.
The following is a quotation of 35 U.S.C. 112 (pre-AIA ), second paragraph:
The specification shall conclude with one or more claims particularly pointing out and distinctly claiming the subject matter which the applicant regards as his invention.
Claims 1-8, 10-11, and 14 are rejected under 35 U.S.C. 112(b) or 35 U.S.C. 112 (pre-AIA ), second paragraph, as being indefinite for failing to particularly point out and distinctly claim the subject matter which the inventor or a joint inventor (or for applications subject to pre-AIA 35 U.S.C. 112, the applicant), regards as the invention.
Claims 1-2 recite the elements “CYP,” “SULT,” and “UGT” without previously defining the acronyms within the claims. While, “CYP,” “SULT,” and “UGT” are defined in the specification, at para. 1, the initial recitation of the acronym within the claims should be accompanied by the fully spelled out words or phrase. Claims 3-8 and 14 are also indefinite due to their dependency upon claim 1.
The term “relative importance” in claims 1, 3-4, and 10 is a relative term which renders the claims indefinite. The term “relative importance” is not defined by the claims, the specification does not provide a standard for ascertaining the requisite degree, and one of ordinary skill in the art would not be reasonably apprised of the scope of the invention. This renders the claims indefinite because it is unclear what the selection of subset of molecular descriptors should be based on. Similarly, claims 2, 5-8, and 14 are rejected due to their dependency upon claim 1.
Claim 7 recites the limitation “wherein the classification model is obtained by training a training dataset.” This renders the claim indefinite because it is unclear how training a training dataset will result in obtaining the classification model. Claim 8 is similarly rejected due to its dependency upon claim 7. For purposes of the present examination, claim 7 will be interpreted to mean that the classification model is obtained by training the model on a training dataset.
Claim 8 recites the limitation "the comparison" in the third limitation. There is insufficient antecedent basis for this limitation in the claim. Additionally, claim 8 recites the limitation "the major vote" in the last limitation. There is insufficient antecedent basis for this limitation in the claim.
Claim 11 recites the limitation "the majority vote" in the last limitation. There is insufficient antecedent basis for this limitation in the claim.
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 7-8 and 14 are rejected under 35 U.S.C. 101 because they do not fall within one of the four enumerated categories of statutory subject matter. Claims 1-6 and 9-12 are rejected under 35 U.S.C. 101 because the claimed invention is directed to a judicial exception (abstract ideas and natural phenomenon) without significantly more. Under MPEP § 2106, subject matter is patent eligible when the claimed invention is to one of the four statutory categories of invention [Step 1], and the claim is not directed to a judicial exception [Step 2A] unless the claim as a whole includes additional limitations amounting to significantly more than the exception [Step 2B].
Step 1
Claims 1-12 describe inventions that are to one of the statutory categories. In Step 1, a claim must fall within one of the four enumerated categories of statutory subject matter (process, machine, manufacture, or composition of matter); a claim falling outside these categories is ineligible without further analysis. See MPEP § 2106.03. Claims 1-6 and 9-12 are properly to one of the four statutory categories because the claimed invention is a method, which falls into the process category [Step 1: Yes].
Claims 7-8 recite a “classification model configured for predicting whether a molecule is an inhibitor of a predetermined enzyme.” A classification model without any structural recitations constitutes software per se, which does not fall into one of the four statutory categories. Therefore, claims 7-8 are not eligible for patent protection at Step 1 because it is directed to non-statutory subject matter.
Claim 14 recites a “non-transitory computer-readable support having stored thereon code instructions.” Computer-readable support is not defined in the instant specification and is not a term known in the art. Under its broadest reasonable interpretation, non-transitory computer-readable support includes software stored in any tangible format, which includes computer code recorded on a piece of paper, which constitutes data per se. A claim whose broadest reasonable interpretation covers both statutory and non-statutory embodiments embraces subject matter that is not eligible for patent protection and therefore is directed to non-statutory subject matter. See MPEP § 2106.03(II). Therefore, claim 14 is not eligible for patent protection because it is directed to non-statutory subject matter when the broadest reasonable interpretation embraces data per se, which is not directed to any of the statutory categories.
Step 2A
Under Step 2A, a claim is directed to a judicial exception if, under the broadest reasonable interpretation, it recites an abstract idea, law of nature, or natural phenomena [Prong One] without the claim as a whole integrating the exception into a practical application [Prong Two]. Abstract ideas include mathematical concepts, mental processes, and certain methods of organizing human activity. Mathematical concepts encompass mathematical relationships, formulas, equations, and mathematical calculations. See MPEP § 2106.04(a)(2)(I). Mental processes involve concepts that can be performed in the human mind or by a human with the aid of pen and paper, such as observations, evaluations, judgments, or opinions. See MPEP § 2106.04(a)(2)(III). Certain methods of organizing human activity include fundamental economic principles, commercial or legal interactions, and managing personal behavior or relationships. See MPEP § 2106.04(a)(2)(II). Laws of nature and natural phenomena, include naturally occurring principles/relations and nature-based products that are naturally occurring or that do not have markedly different characteristics compared to what occurs in nature. See MPEP § 2106.04(b)-(c).
Prong One
A claim recites a judicial exception when it sets forth or describes a law of nature, natural phenomenon, or abstract idea. Claims 1-6 and 9-12 recite abstract ideas that fall into the groupings of mathematical concepts and mental processes.
Claims 1-6 recite the following limitations, which describe abstract ideas within the mathematical concepts and/or mental processes groupings:
Claim 1 recites selecting, from an initial set of molecular descriptors comprising physicochemical molecular descriptors and at least one binding energy on at least one conformation of the determined CYP, SULT or UGT enzyme, a subset of molecular descriptors, based on the relative importance of the descriptors in predicting the inhibiting character of a molecule, and
Claim 1 recites performing a supervised training, over the training dataset, of a classification model configured to receive as input a vector formed of the subset of molecular descriptors computed on a molecule, and to output an indication of the inhibiting character of the molecule on the determined CYP, SULT or UGT enzyme.
Claim 2 recites wherein the determined CYP, SULT or UGT enzyme is selected from the group consisting of: CYP 2C9, CYP 2D6, SULT 1A1, SULT 1A3, and UGT 1A1.
Claim 3 recites wherein selecting the subset of descriptors based on their relative importance comprises training a plurality of random forest models on the training dataset, computing a Gini importance index of all descriptors of the set, and selecting the molecular descriptors having highest Gini importance.
Claim 4 recites wherein determining the number of descriptors to select based on their relative importance comprises computing an average balanced accuracy of a plurality of random forest models with multiple sets of descriptors having a varying number of descriptors, and selecting the number of descriptors maximizing the average balanced accuracy.
Claim 5 recites prior to the step of selecting, a step of removing, from the initial set of molecular descriptors: highly correlated descriptors; descriptors having missing or infinite values on data of the training dataset; and descriptors having a variance below a determined threshold over the training dataset.
Claim 6 recites wherein the classification model is a random forest model or a Support Vector Machine model.
The claim 1 limitation of selecting a subset of molecular descriptors constitutes an abstract idea within the mathematical concepts and mental processes groupings. This limitation is narrowed by the limitation of claim 3 because it specifies that the subset of descriptors is selected by computing a Gini importance index, which constitutes a mathematical concept. This is further narrowed by the limitation of claim 4 because it specifies that the number of descriptors is determined by computing an average balanced accuracy, which constitutes a mathematical concept. The claim 1 limitation of performing a supervised training involves utilizing optimization, calculus, linear algebra, and probability theory, which constitutes an abstract idea within the mathematical concepts grouping.
The limitation of claim 2 narrows the abstract ideas of claim 1 by specifying the enzyme. The limitation of claim 5 involves removing descriptors from the initial set based on correlations, missing or infinite values, and values below a threshold, which constitutes an abstract idea within the mathematical concepts and mental processes groupings. The limitation of claim 6 narrows the abstract ideas of claim 1 by specifying the type of classification model.
Claims 9-12 recite the following limitations, which describe abstract ideas within the mathematical concepts and/or mental processes groupings:
Claim 9 recites computing a set of molecular descriptors of the candidate molecule and at least one binding energy of the candidate molecule on at least one conformation of the enzyme,
Claim 9 recites providing the computed set of molecular descriptors and the at least one computed binding energy to a classification model trained to output, from the set of molecular descriptors and said at least one binding energy of the candidate molecule on a conformation of the enzyme, an indication output about whether said candidate molecule is an inhibitor or non-inhibitor of the enzyme
Claim 10 recites training the classification model by selecting, from an initial set of molecular descriptors comprising physicochemical molecular descriptors and at least one binding energy on at least one conformation of the enzyme, a subset of molecular descriptors based on the relative importance of the molecular descriptors in predicting the inhibiting character of a molecule, and performing a supervised training using a training dataset of a classification model configured to receive as input a vector formed of the subset of molecular descriptors, and to output an indication of whether the candidate molecule is an inhibitor or non-inhibitor of the enzyme.
Claim 11 recites receiving an indication from each classifier as to whether the candidate module is an inhibitor or non-inhibitor of the predetermined enzyme,
Claim 11 recites computing, for a plurality of conformations of the enzyme, a binding energy of the candidate molecule with each conformation of the enzyme,
Claim 11 recites comparing the lowest computed binding energy with two thresholds and inferring, from said comparison, a third indication, and
Claim 11 recites determining whether the candidate molecule is an inhibitor or non- inhibitor of the enzyme according to the majority vote over the three indications.
Claim 12 recites wherein the candidate molecule is a candidate drug or a xenobiotic.
The limitation of claim 9 of computing a set of molecular descriptors is an abstract idea within the mathematical concepts and mental processes groupings. The limitation of claim 9 of providing the set of descriptors to output an indication of inhibition status is an abstract idea within the mathematical concepts and mental processes groupings. The limitation of claim 10 of selecting a subset of molecular descriptors constitutes an abstract idea within the mathematical concepts and mental processes groupings. The limitation of claim 10 of performing a supervised training involves utilizing optimization, calculus, linear algebra, and probability theory, which constitutes an abstract idea within the mathematical concepts grouping. The limitation of claim 11 of receiving an indication of inhibition status from the models involves the execution of mathematical functions to determine the inhibition status, which constitutes an abstract idea within the mathematical concepts and mental processes groupings. The limitation of claim 11 of computing a binding energy is a mathematical concept that could be performed mentally or with the aid of pen and paper, which constitutes an abstract idea. The limitation of claim 11 of comparing the lowest computed binding energy with two thresholds is an abstract idea within the mathematical concepts and mental processes groupings. The limitation of claim 11 of determining whether the candidate molecule is an inhibitor or non-inhibitor is an abstract idea within the mathematical concepts and mental processes groupings. The limitation of claim 12 narrows the abstract ideas of claim 9 by specifying the candidate molecule as a candidate drug or a xenobiotic.
Therefore, claims 1-6 and 9-12 recite abstract ideas – namely mathematical concepts and mental processes [Step 2A, Prong One: Yes].
Prong Two
Claims 1-6 and 9-12 as a whole do not integrate the recited judicial exception into a practical application. A claim that recites a judicial exception [Prong One] is deemed to be directed to a judicial exception [Step 2A] unless the claim as a whole contains additional elements that integrate the exception into a practical application [Prong Two]. 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, beyond generally linking the use of the judicial exception to a particular technological environment, such that the claim as a whole is more than a drafting effort designed to monopolize the exception. See MPEP §§ 2106.04(d) and 2106.05(e). A claim does not integrate a judicial exception into a practical application by reciting insignificant extra-solution activity, generally linking the exception to a particular technological environment or field of use, merely reciting to apply the exception, merely including instructions to implement an abstract idea on a computer, or merely using a computer as a tool to perform an abstract idea. See MPEP § 2106.04(d)(I). Insignificant extra-solution activities are nominal or tangential additions to a claim that are incidental to the primary process or product, including both pre-solution and post-solution activity (e.g. pre-solution data gathering for use in a process). If integrated into a practical application, the claim is eligible; otherwise, it is directed to the judicial exception, necessitating further analysis at Step 2B.
Claims 9 and 11 recite the following limitations, which are additional elements:
Claim 9 recites receiving the indication output by the classification model about whether said candidate molecule is an inhibitor or non-inhibitor of the enzyme.
Claim 11 recites providing the set of molecular descriptors and each computed binding energy to a first classifier formed by a random forest model and a second classifier formed by a Support Vector Machine model
The limitation of claim 9 is an insignificant post-solution activity step of receiving data, which does not integrate the judicial exceptions into a practical application. See MPEP § 2106.05(g). The limitation of claim 11 recites providing input to classification models, which is a field of use limitation equivalent to the words “apply it” that does not integrate the judicial exceptions into a practical application. See MPEP §§ 2106.05(f) & (h). Claims 1-6, 10, and 12 do not include any additional elements.
The claims as a whole merely recite insignificant extra-solution activities and abstract ideas implemented on generic computer components without meaningful limitations that tie it to a specific technological improvement. Therefore, claims 1-6 and 9-12 do not contain additional elements that integrate the recited abstract ideas into a practical application [Step 2A, Prong Two: No].
Step 2B
Claims 1-6 and 9-12 do not include additional elements, whether considered individually or in combination, that are sufficient to amount to significantly more than the judicial exception itself. Under Step 2B, the claim is analyzed to determine whether there are any additional elements that, individually or in combination, constitute an “inventive concept" sufficient to ensure that the claim, as a whole, amounts to significantly more than the judicial exception itself. See MPEP § 2106.05; and Alice Corp. Pty. Ltd. v. CLS Bank Int'l, 573 U.S. 208, 217-18, 110 USPQ2d 1976, 1981 (2014).
Claims 9 and 11 recite the following limitations, which are additional elements:
Claim 9 recites receiving the indication output by the classification model about whether said candidate molecule is an inhibitor or non-inhibitor of the enzyme.
Claim 11 recites providing the set of molecular descriptors and each computed binding energy to a first classifier formed by a random forest model and a second classifier formed by a Support Vector Machine model
The limitation of claim 9 is a conventional insignificant post-solution activity step of receiving data, which does not add significantly more than the judicial exceptions. See MPEP §§ 2106.05(d) & (g); and OIP Techs., Inc., v. Amazon.com, Inc., 788 F.3d 1359, 1363, 115 USPQ2d 1090, 1093 (Fed. Cir. 2015). The limitation of claim 11 recites providing input to classification models, which is a field of use limitation equivalent to the words “apply it” that does not add significantly more than the judicial exceptions themselves. See MPEP §§ 2106.05(f) & (h); and Robin Winter et al., Learning continuous and data-driven molecular descriptors by translating equivalent chemical representations, 10(6) Chem. Sci. 1692, Abstract (19 November 2018).
Overall, claims 1-6 and 9-12 amount to no more than insignificant extra-solution activities and implementing the abstract ideas on conventional computers in a routine way. The claims do not include additional elements that are sufficient to amount to significantly more than the judicial exception itself because the claims recite additional elements that equate to insignificant extra-solution activity and mere instructions to apply the recited abstract ideas in a particular field. Therefore, claims 1-6 and 9-12 are rejected for failing to set forth patent eligible subject matter under 35 U.S.C. 101 because the claimed invention recites abstract ideas [Step 2A, Prong One: Yes] and the additional elements do not integrate the judicial exception into a practical application [Step 2A, Prong Two: No] and do not amount to claiming significantly more than the recited exception [Step 2B: No].
Claim Rejections - 35 USC § 102
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 the appropriate paragraphs of 35 U.S.C. 102 that form the basis for the rejections under this section made in this Office action:
A person shall be entitled to a patent unless –
(a)(1) the claimed invention was patented, described in a printed publication, or in public use, on sale, or otherwise available to the public before the effective filing date of the claimed invention.
Claim 9 is rejected under 35 U.S.C. 102(a)(1) as being anticipated by Virginie Y. Martiny et al., Integrated structure- and ligand-based in silico approach to predict inhibition of cytochrome P450 2D6, 31(24) Bioinformatics 3930-37 (26 August 2015) (hereinafter “Martiny”).
Regarding claim 9, Martiny discloses a method for predicting inhibition of CYP2D6. At 3932 col.2 para.4 (a method for predicting whether a candidate molecule is an inhibitor of a enzyme). Martiny discloses computing molecular descriptors and the protein–ligand binding energies calculated on the best performing conformations. At 3932 col.2 para.3 (computing a set of molecular descriptors of the candidate molecule and at least one binding energy of the candidate molecule on at least one conformation of the enzyme). Martiny uses the molecular descriptors and the binding energies as input for classification models trained to output a prediction of CYP2D6 inhibition. At 3932 col.2 paras.3-4 (providing the computed set of molecular descriptors and the at least one computed binding energy to a classification model trained to output, from the set of molecular descriptors and said at least one binding energy of the candidate molecule on a conformation of the enzyme, an indication output about whether said candidate molecule is an inhibitor or non-inhibitor of the enzyme). While Martiny does not explicitly teach receiving the indication output by the classification model about whether said candidate molecule is an inhibitor or non-inhibitor of the enzyme, Martiny discloses that the models predict CYP2D6 inhibition with an accuracy of 75%. At 3936 col.2 para.1. Thus, inherit in the disclosure of Martiny is receiving the model output, which indicates the inhibition status of CYP2D6.
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.
Claim 12 is rejected under 35 U.S.C. 103 as being unpatentable over Martiny.
Regarding claim 12, Martiny teaches that the enzyme CYP is responsible for the metabolism of drugs and xenobiotic substances. At 3931 col.1 para.1. While Martiny does not explicitly disclose the candidate molecule being a candidate drug or a xenobiotic, a person having ordinary skill in the art would understand that the molecule being labeled as an inhibitor or non-inhibitor of the CYP enzyme should be a candidate drug or a xenobiotic. One of ordinary skill in the art would reasonably expect success in assessing the inhibition status of a candidate drug or a xenobiotic because CYP is responsible for the metabolism of drugs and xenobiotic substances.
Claims 1-2, 6-7, 10, and 14 are rejected under 35 U.S.C. 103 as being unpatentable over Martiny in view of Gaspar Cano et al., Automatic selection of molecular descriptors using random forest: Application to drug discovery, 72 Expert Syst. Appl. 151-59 (6 December 2016) (hereinafter “Cano”), as evidenced by C3.ai., Infrastructure: Machine Learning Hardware Requirements, (15 May 2021) (hereinafter “C3.ai.”).
Regarding claim 1, Martiny discloses a method of training a model to predict inhibition of CYP2D6. At 3932 col.2 paras.2-4 (a method for training a model for predicting inhibitors of a determined CYP, SULT or UGT enzyme). Martiny built a balanced training dataset containing molecules known to be an inhibitor and molecules known to be a non-inhibitor of CYP2D6. At 3932 col.2 para.2 (a training dataset comprising a plurality of molecules known as being an inhibitor or non-inhibitor of the determined CYP, SULT or UGT enzyme). Martiny initially explores binding site flexibility of two CYP2D6 structures and determines binding energies. At 3931 col.2 para.4 – 3932 col.1 para.1 (from an initial set of molecular descriptors comprising physicochemical molecular descriptors and at least one binding energy on at least one conformation of the determined CYP, SULT or UGT enzyme; instant spec p.10 lns.1-5: physicochemical molecular descriptors, representing features of the molecules such as its … flexibility). To obtain the input for the model, Martiny performs principal component analysis to reduce the dimensionality and selects extended connectivity fingerprints and the protein–ligand binding energies calculated on the best performing MD receptor conformations as molecular descriptors. At 3932 col.2 paras.2-3 (selecting … a subset of molecular descriptors). Using the constructed dataset, Martiny trains three classification models, including a support vector machine (SVM), a random forest-based predictor, and a NaiveBayesian predictor. At 3932 col.2 paras.2-3 (performing a supervised training, over the training dataset, of a classification model configured to receive as input a vector formed of the subset of molecular descriptors computed on a molecule). Martiny discloses that the models are able to predict whether CYP2D6 is inhibited. At 3932 col.2 para.4 (to output an indication of the inhibiting character of the molecule on the determined CYP, SULT or UGT enzyme).
Martiny fails to explicitly teach a training device comprising a computer and a memory. However, Martiny’s disclosure relates to machine learning models, which necessarily
involve a computer and a memory. See C3.ai, §§ Processors: CPUs, GPUs, TPUs, and FPGAs - Memory and Storage. Additionally, Martiny fails to teach selecting the subset of molecular descriptors based on the relative importance of the descriptors in predicting the inhibiting character of a molecule. However, Cano discloses using a Random Forest (RF)-based approach to select molecular descriptors for ligands of enzymes, and classification using the automatically selected feature subset. Abstract. Cano teaches that in RF, a ranking of the contribution of each variable is determined to predict the output variable, establishing a relative importance between them. At 154 col.1 para.2. Cano notes that RF-based feature selection can have the following benefits: reduction of the data to be processed; reduction of features, reducing the cost of continued storage; improved performance, improved processing speed can lead to an improvement in prediction accuracy; improved display, improved representation helps the understanding of the problem; reduced training time, smaller data subset decreases training time; and reduction of noise in the data, removing irrelevant or redundant features. At 152 col.1 para.1.
A person having ordinary skill in the art would be motivated to combine the teachings of Cano with the teachings of Martiny because RF-based feature selection offers numerous benefits, including reduction of time and cost, and improvement of performance. One of ordinary skill in the art would reasonably expect the combination to result in an improved method of enzyme inhibitor prediction because the dimensionality reduction via principal component analysis disclosed by Martiny can be replaced with dimensionality reduction via RF-based feature selection disclosed by Cano. Some teaching, suggestion, or motivation in the prior art that would have led one of ordinary skill to modify the prior art reference or to combine prior art reference teachings to arrive at the claimed invention is likely to be obvious. See KSR International Co. v. Teleflex Inc., 550 U.S. 398, 415-421, USPQ2d 1385, 1395 – 97 (2007); and MPEP § 2143, G.
Regarding claim 2, Martiny discloses a method of training a model to predict inhibition of CYP2D6. At 3932 col.2 paras.2-4 (the method according to claim 1, wherein the determined CYP, SULT or UGT enzyme is selected from the group consisting of: ).
Regarding claim 6, Martiny trains three classification models, including a support vector machine (SVM), a random forest-based predictor, and a NaiveBayesian predictor. At 3932 col.2 paras.2-3 (the method according to claim 1, wherein the classification model is a random forest model or a Support Vector Machine model).
Regarding claim 7, Martiny discloses a classification model to predict inhibition of CYP2D6 by training the model using the method discussed above (see claim 1 rejection above). At 3932 col.2 paras.2-4 (a classification model configured for predicting whether a molecule is an inhibitor of a predetermined enzyme, wherein the classification model is obtained by training a training dataset in accordance with the method of claim 1).
Regarding claim 10, Martiny discloses the method of claim 9 (see 102 rejection above). To train the models, Martiny initially explores binding site flexibility of two CYP2D6 structures and determines binding energies. At 3931 col.2 para.4 – 3932 col.1 para.1 (from an initial set of molecular descriptors comprising physicochemical molecular descriptors and at least one binding energy on at least one conformation of the enzyme; instant spec p.10 lns.1-5: physicochemical molecular descriptors, representing features of the molecules such as its … flexibility). To obtain the input for the model, Martiny performs principal component analysis to reduce the dimensionality and selects extended connectivity fingerprints and the protein–ligand binding energies calculated on the best performing MD receptor conformations as molecular descriptors. At 3932 col.2 paras.2-3 (selecting … a subset of molecular descriptors). Using the constructed dataset, Martiny trains three classification models, including a support vector machine (SVM), a random forest-based predictor, and a NaiveBayesian predictor. At 3932 col.2 paras.2-3 (performing a supervised training using a training dataset of a classification model configured to receive as input a vector formed of the subset of molecular descriptors). Martiny discloses that the models are able to predict whether CYP2D6 is inhibited. At 3932 col.2 para.4 (to output an indication of whether the candidate molecule is an inhibitor or non-inhibitor of the enzyme). Cano teaches that in RF-based feature selection, a ranking of the contribution of each variable is determined to predict the output variable, establishing a relative importance between them. At 154 col.1 para.2 (based on the relative importance of the molecular descriptors in predicting the inhibiting character of a molecule).
Regarding claim 14, Martiny’s disclosure relates to machine learning models, which necessarily involves a computer-readable memory storing instructions. See C3.ai, § Memory and Storage (a non-transitory computer-readable support having stored thereon code instructions which, when executed by a computer, cause the computer to carry out the method according to claim 1).
Claim 3 is rejected under 35 U.S.C. 103 as being unpatentable over Martiny and Cano as applied to claims 1-2, 6-7, 10, and 14 above, and further in view of Philipp Probst et al., Hyperparameters and Tuning Strategies for Random Forest, 9 WIREs Data Mining and Knowledge Discovery (26 February 2019) (hereinafter “Probst”).
Regarding claim 3, Martiny discloses using the constructed dataset to train three classification models, including a support vector machine (SVM), a random forest-based predictor, and a NaiveBayesian predictor. At 3932 col.2 paras.2-3 (the method according to claim 1, wherein selecting the subset of descriptors based on their relative importance comprises training ). Cano discloses that RF calculates the Mean Decrease Gini from the Gini index of the features, which is used to select the features having the highest importance. At 154 col.1 paras.2-4 (computing a Gini importance index of all descriptors of the set, and selecting the molecular descriptors having highest Gini importance). While Cano only discloses using a single random forest model, Probst teaches training several random forests to assess the stability of the variable importance estimates. At 6 para.2. Probst discloses that training more trees in the model results in a more stable and reliable estimate of variable importance. At 6 para.2.
A person having ordinary skill in the art would be motivated to combine the teachings of Probst with the teachings of Martiny and Cano because training a plurality of RF models allows for one to assess the stability and reliability of the variable importance estimates. One of ordinary skill in the art would reasonably expect the combination to result in an improved method of enzyme inhibitor prediction because employing Probst’s plurality of models results in a more reliable final model for variable selection. Some teaching, suggestion, or motivation in the prior art that would have led one of ordinary skill to modify the prior art reference or to combine prior art reference teachings to arrive at the claimed invention is likely to be obvious. See KSR International Co. v. Teleflex Inc., 550 U.S. 398, 415-421, USPQ2d 1385, 1395 – 97 (2007); and MPEP § 2143, G.
Claims 4-5 are rejected under 35 U.S.C. 103 as being unpatentable over Martiny, Cano, and Probst as applied to claim 3 above, and further in view of Robin Genuer et al., Variable selection using random forests, 31 Pattern Recognit. Lett. 2225-36 (21 March 2010) (hereinafter “Genuer”).
Regarding claim 4, Martiny, Cano, and Probst fail to disclose determining the number of descriptors to select by computing an average balanced accuracy of a plurality of random forest models with multiple sets of descriptors having a varying number of descriptors, and selecting the number of descriptors maximizing the average balanced accuracy. However, Genuer discloses performing RF-based variable selection and assesses the sensitivity to the sample size n and the number of variables p. At 2227 col.2 para.2. Genuer uses a RF model and performs 50 runs, varying the sample size and the number of variables. At 2228 col.1 paras.2-4 (a plurality of random forest models with multiple sets of descriptors having a varying number of descriptors). Genuer ranks variables by averaging the variable importance resulting from the 50 runs. At 2231 col.2 para.5 (computing an average balanced accuracy). To determine the number of variables, Genuer selects the variables of the model leading to the smallest out-of-bag (OOB) error. At 2232 col.2 para.4 (selecting the number of descriptors maximizing the average balanced accuracy).
The combination of Martiny, Cano, and Probst discloses a base method of enzyme inhibitor prediction using a plurality of random forest models for RF-based variable selection. Genuer discloses a technique for determining the number of variables to select in RF-based variable selection. A person having ordinary skill in the art would recognize that applying Genuer’s technique to the combination of Martiny, Cano, and Probst would predictably yield an improved system because selecting the number of variables based on an average of RF runs enhances model stability and reliability. Therefore, it would be obvious to one of ordinary skill in the art to use Genuer’s average variable selection technique in the combination of Martiny, Cano, and Probst. Applying a known technique to a known device (method, or product) ready for improvement to yield predictable results is likely to be obvious. See KSR International Co. v. Teleflex Inc., 550 U.S. 398, 415-421, USPQ2d 1385, 1395 – 97 (2007); and MPEP § 2143, D.
Regarding claim 5, Probst discloses using a dataset that has no missing values. At 11 para.5. While Probst does not explicitly disclose removing descriptors having missing values, one of ordinary skill in the art would understand from this disclosure that descriptors with missing values should be removed from the training dataset before selecting the subset of molecular descriptors. Genuer discloses that before selecting the final variables, a threshold for importance is established and variables falling below this threshold are removed. At 2232 col.1 para.1 (prior to the step of selecting, a step of removing, from the initial set of molecular descriptors: descriptors having a variance below a determined threshold over the training dataset). Genuer also teaches that the error rate of RF models increases when highly correlated variables are present in the dataset. At 2233 col.1 para.2. While Genuer does not disclose removing highly correlated variables from the dataset, one of ordinary skill in the art would understand that highly correlated variables should be removed prior to selecting the subset of molecular descriptors because the removal will result in a model with a lower error rate.
Claims 8 and 11 are rejected under 35 U.S.C. 103 as being unpatentable over Martiny and Cano as applied to claims 1-2, 6-7, 10, and 14 above, and further in view of Oleg V. Stroganov et al., Lead Finder: An Approach To Improve Accuracy of Protein−Ligand Docking, Binding Energy Estimation, and Virtual Screening, 48(12) J. Chem. Inf. Model. 2371-85 (13 November 2008) (hereinafter “Stroganov”), Xuelin Zhou et al., Molecular docking and enzyme kinetic studies of dihydrotanshinone on metabolism of a model CYP2D6 probe substrate in human liver microsomes, 19(7) Phytomedicine 648-57 (15 May 2012) (hereinafter “Zhou”), and Sanchita Mangale, Voting Classifier, Medium (18 May 2019) (hereinafter “Mangale”).
Regarding claim 8, Martiny discloses training three classification models, including a support vector machine (SVM) and a RF-based predictor. At 3932 col.2 paras.2-3 (a first classifier formed by a random forest model trained according to 7; a second classifier formed a Support Vector Machine model trained according to 7). Martiny discloses computing the binding energies for multiple conformations of CYP2D6, and selecting the confirmation with the lowest binding energy. At 3932 col.1 para.1 (the lowest binding energy computed for a plurality of conformations of the predetermined enzyme). Martiny and Cano fail to teach a third classifier indicating whether a molecule is an inhibitor of the predetermined enzyme based on a comparison of the lowest binding energy with at least one threshold; and the output of the model being the major vote over the three classifiers. However, Stroganov discloses screening potential enzyme inhibitors based on binding energy. Abstract. Additionally, Zhou teaches that the binding energy of CYP2D6 inhibitors can range from -5.3 kcal/mol to -9.1 kcal/mol. Table 3. Moreover, Mangale teaches that predictions from multiple machine learning models can be combined in an ensemble learning method by implementing a voting classifier. At paras. 1-3. Mangale discloses that a hard voting classifier produces an output based on the majority vote between the machine learning models. At paras.5-6. Mangale notes that an ensemble learning method with a voting classifier can lower the error rate and reduce over-fitting. At para.2.
Martiny and Cano disclose a base method of enzyme inhibitor prediction. Stroganov discloses screening potential enzyme inhibitors based on binding energy, and Zhou teaches the typical range of binding energies for CYP2D6 inhibitors. A person having ordinary skill in the art would recognize that implementing the techniques of Stroganov by screening for binding energies within the typical range taught by Zhou would result in an improved method of enzyme inhibitor prediction because the additional screening reduces the number of candidate molecules. Additionally, Mangale teaches combining multiple machine learning models via a voting classifier. A person having ordinary skill in the art would recognize that applying the voting classifier technique of Mangale to the combination of Martiny, Cano, Stroganov, and Zhou would result in an improved method of enzyme inhibitor prediction because the voting classifier incorporates the output of each model to reduce overall error rate and over-fitting. Applying a known technique to a known device (method, or product) ready for improvement to yield predictable results is likely to be obvious. See KSR International Co. v. Teleflex Inc., 550 U.S. 398, 415-421, USPQ2d 1385, 1395 – 97 (2007); and MPEP § 2143, D.
Regarding claim 11, Martiny discloses providing the set of selected molecular descriptors and binding energies to a random forest model and a SVM model. At 3932 col.2 para.3 (providing the set of molecular descriptors and each computed binding energy to a first classifier formed by a random forest model and a second classifier formed by a Support Vector Machine model). While Martiny does not explicitly teach receiving an indication from each classifier as to whether the candidate module is an inhibitor or non-inhibitor of the predetermined enzyme, Martiny discloses that the models predict CYP2D6 inhibition with an accuracy of 75%. At 3936 col.2 para.1. Thus, inherit in the disclosure of Martiny is receiving the models output, which indicates the inhibition status of CYP2D6. Martiny discloses computing the binding energies for multiple conformations of CYP2D6, and selecting the confirmation with the lowest binding energy. At 3932 col.1 para.1 (computing, for a plurality of conformations of the enzyme, a binding energy of the candidate molecule with each conformation of the enzyme). Stroganov discloses screening potential enzyme inhibitors based on binding energy. Abstract. Zhou teaches that the binding energy of CYP2D6 inhibitors can range from -5.3 kcal/mol to -9.1 kcal/mol. Table 3. In implementing Stroganov’s binding energy-based screening, one of ordinary skill in the art would understand that the candidate molecules should be screened based on thresholds informed by the teachings of Zhou, such that molecules with a binding energy higher than -5.3 kcal/mol or lower than -9.1 kcal/mol are identified as non-inhibitors (comparing the lowest computed binding energy with two thresholds and inferring, from said comparison, a third indication). Finally, Mangale discloses that a hard voting classifier produces an output based on the majority vote between the machine learning models. At paras.5-6 (determining whether the candidate molecule is an inhibitor or non- inhibitor of the enzyme according to the majority vote over the three indications).
Conclusion
Any inquiry concerning this communication or earlier communications from the examiner should be directed to Emily A Darrigrand whose telephone number is (571) 272-1098. The examiner can normally be reached Monday-Thursday 7:00AM-4:00PM.
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, Larry Riggs, can be reached at (571) 270-3062. 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.
/E.A.D./Examiner, Art Unit 1686
/LARRY D RIGGS II/Supervisory Patent Examiner, Art Unit 1686
Prosecution Insights