DETAILED ACTION
Applicant’s response, 03 Dec. 2025 has been fully considered. The following rejections and/or objections are either reiterated or newly applied. They constitute the complete set presently being applied to the instant application.
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 Claims
Claims 10, 17, 20-24, 26-45, 47, 49-63, 65-66, 68-76, and 78-126 are cancelled.
Claims 127-139 are newly added.
Claims 1-9, 11-16, 18-19, 25, 46, 48, 64, 67, 77, and 127-139 are pending.
Claims 3-4, 64, and 77 are withdrawn from further consideration pursuant to 37 CFR 1.142(b) as being drawn to a nonelected invention or species, there being no allowable generic or linking claim. Election was made without traverse in the reply filed on 12 March 2024.
Claims 1-2, 5-9, 11-16, 18-19, 25, 46, 48, 67, and 127-139 are rejected.
Priority
The following is previously recited:
Applicant’s claim for the benefit of a prior-filed application, US. Provisional App. No. 63/298,012 filed 10 Jan. 2022 and U.S. Provisional App. No. 63/307,957 filed 08 Feb. 2022 under 35 U.S.C. 119(e) or under 35 U.S.C. 120, 121, 365(c), or 386(c) is acknowledged.
Applicant has not complied with one or more conditions for receiving the benefit of an earlier filing date as follows:
The later-filed application must be an application for a patent for an invention which is also disclosed in the prior application (the parent or original nonprovisional application or provisional application). The disclosure of the invention in the parent application and in the later-filed application must be sufficient to comply with the requirements of 35 U.S.C. 112(a) or the first paragraph of pre-AIA 35 U.S.C. 112, except for the best mode requirement. See Transco Products, Inc. v. Performance Contracting, Inc., 38 F.3d 551, 32 USPQ2d 1077 (Fed. Cir. 1994).
The disclosure of the prior-filed application, Application No. 63/298,012, fails to provide adequate support or enablement in the manner provided by 35 U.S.C. 112(a) or pre-AIA 35 U.S.C. 112, first paragraph for one or more claims of this application. The specification of app. ‘012 does not provide support for multi-objective search algorithms or a prescriptive model.
Therefore, the effective filing date of the claimed invention is 08 Feb. 2022.
Information Disclosure Statement
The information disclosure statement (IDS) submitted on 19 Feb. 2026 is in compliance with the provisions of 37 CFR 1.97. Accordingly, the list of cited references was considered by the examiner.
Claim Objections
Claim 1 is objected to because of the following informalities:
Claim 1 recites “(a) obtaining information comprising one or more substrates at each iteration that is after an initial iteration in the multiple iterations: generating…”, which is grammatically incorrect and nonsensical. Claim 1 should be amended to recite “(a) obtaining information…in the multiple iterations, wherein each iteration after the initial iteration comprises: generating…”, to increase clarity that the following steps (before step (b)) are substeps of (a).
Appropriate correction is required.
Claim Interpretation
Claims 128, 133, and 137 recite “wherein the activity….comprises a novel activity for the target/candidate protein”. In light of Applicant’s specification at para. [0342], “novel activity” of a protein is interpreted to mean an activity for which the wild-type form of the protein has no activity.
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.
Claims 16, 18-19, 25, 46 and 132-135 rejected under 35 U.S.C. 112(a) 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 at the time the application was filed, had possession of the claimed invention. This rejection is newly recited and necessitated by claim amendment.
Claim 16, and claims dependent therefrom, recite “…processing at least the set of physicochemical attributes by the trained machine learning model to generate the predicted activity of the candidate protein on multiple design objectives, wherein the multiple design objectives comprise activity and stability”. Accordingly, claim 16 requires using a trained machine learning model to process physicochemical attributes to generate a predicted activity of the candidate protein on design objectives including activity and stability.
Dependent claim 132 further recites “the multiple design objectives comprise enzymatic activity, selectivity, stability, toxicity, size, novelty, or any combination thereof”.
However, Applicant’s specification does not provide support for “multiple design objectives” comprising “activity and stability” for the trained machine learning model. Instead, Applicant’s specification at para. [0119], [0121], [0123], [0169] discloses that candidate proteins (i.e. candidate amino acid sequences) are proposed using a prescriptive model that comprises a multi-objective search and optimization algorithm, and at para. [0069] and [0211] that the candidate protein obtained from the prescriptive model is expected to have an optimal enzymatic quality in relation to multiple objectives, and the multiple objectives comprise enzymatic activity, selectivity, stability, toxicity, size, novelty, or a combination thereof. Applicant’s specification does not discuss using the multiple-objectives and/or a multi-objective search algorithm with activity and stability with respect to the trained machine learning model (i.e. the predictive model).
For the reasons discussed above, the specification does not provide a sufficient disclosure of the limitation above of claim 16, and claims dependent thereform, to demonstrate to one of ordinary skill in the art that the inventor possessed the invention at the time the application was filed. For more information regarding the written description requirement, see MPEP §2161.01- §2163.07(b).
Claim Rejections - 35 USC § 112(b)
The rejection of claim 11 under 35 U.S.C. 112(b) in the Office action mailed 04 June 2025 has been withdrawn in view of claim amendments received 03 Dec. 2025.
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.
Claims 1-2, 5-9, 11-16, 18-19, 25, 46, 48, 67, and 127-139 are rejected under 35 U.S.C. 112(b) as being indefinite for failing to particularly point out and distinctly claim the subject matter which the inventor or a joint inventor regards as the invention. This rejection is newly recited and necessitated by claim amendments.
Claim 1 is indefinite for recitation of “generating a plurality of candidate amino acid sequences…using a process…that comprises (i) an actual activity that has been measured…,a and wherein each candidate amino acid sequence represents a candidate enzyme design for the target protein”. Claim 1 later recites “ for each of the plurality of candidate amino acid sequences, generating a respective predicted activity of the target protein”, and repeatedly refers to “activity” and “target protein” (the terms “enzymatic” and “enzyme” have been removed across claim 1). Accordingly, claim 1 uses the broad terms of a “target protein” having an “activity”, and also states each candidate amino acid sequence for the target protein represents a “candidate enzyme design”. A broad range or limitation together with a narrow range or limitation that falls within the broad range or limitation (in the same claim) may be considered indefinite if the resulting claim does not clearly set forth the metes and bounds of the patent protection desired. See MPEP § 2173.05(c). In the present instance, claim 1 recites a “target protein” and the claim also recites the amino sequence for this protein represents “a candidate enzyme design” which is the narrower statement of the range/limitation. The claim(s) are considered indefinite because there is a question or doubt as to whether the feature introduced by such narrower language is (a) merely exemplary of the remainder of the claim, and therefore not required, or (b) a required feature of the claims. As a result, it is not clear if the candidate amino acid sequence can be any protein with any activity (as suggested by a majority of the claim), or of the limitation “each candidate amino acid sequence represents a candidate enzyme design for the target protein” requires that the protein is an enzyme and the activity is an enzyme activity. For purpose of examination, the protein will be interpreted to be an enzyme with enzymatic activity.
Claim 1 is indefinite for recitation of “selecting….candidate amino acid sequences that are in the subset have higher levels of respective predicted enzymatic activities”. There is insufficient antecedent basis for “respective predicted enzymatic activities” of the candidate amino acid sequences because claim 1 previously recites “generating…a respective predicted activity of the target protein…“ but does not require the predicted activity is an enzymatic activity. As a result, for the same reasons discussed above, it is unclear if the activity of the target protein can be any activity or if this must be an enzymatic activity (and thus the protein must be an enzyme). For purpose of examination, the activity is interpreted to be enzymatic activity.
Claim 1, and claims dependent therefrom, are indefinite for recitation of “…(b)… the target enzyme”. There is insufficient antecedent basis for this limitation in the claim because claim 1 only previously recites a “target protein”. For the same reasons discussed above, it is not clear if the target protein is required to be a target enzyme or if the protein can be any type of protein. For purpose of examination, the target protein is interpreted to be a target enzyme.
Claim 16, and claims dependent therefrom, are indefinite for recitation of “…processing at least the set of physicochemical attributes by the trained machine learning model to generate the predicted activity of the candidate protein on multiple design objectives, wherein the multiple design objectives comprise activity and stability”. Given the machine learning model of claim 16 only takes the set of physicochemical attributes as input to predict an activity, it is unclear in what way the limitation “on multiple design objectives” is intended to affect how the machine learning model processes information, if the limitation is intended to specify how the machine learning model was previously trained, or if the design objectives is merely the intended result of the generated predicted activity. For example, it is not clear in what way a machine learning model predicts “activity” using a multiple design objective of “activity” (is “activity” the prediction or being used as an objective to design something else). As discussed above under 35 U.S.C. 112(a), Applicant’s specification only provide support for generating the candidate amino acid sequences using a prescriptive model that uses a multi-objective search algorithm (i.e. the activity is used as a design objective in generating an amino acid sequence), and thus does not serve to clarify the metes and bounds of the claim. Clarification is requested. For purposes of applying prior art, given the claim only processes the attributes to generate a predicted activity of a candidate protein, the phrase “on multiple design objectives…comprise activity and stability” is interpreted to define the intended use or result (i.e. the objective) of the prediction.
Claim 48, and claims dependent therefrom, for recitation of “(e) measuring an actual activity on the substrate of the expressed protein…”. Claim 48 previously recites “…expressing the subset…to generate one or more expressed proteins and exposing the one or more expressed proteins to the substrate”, and therefore it is unclear if “the expressed protein” is referring to each of “the one or more expressed proteins” or a particular expressed protein of the one or more expressed proteins. If Applicant intends for “the expressed protein” to refer to a particular protein, it is further unclear which protein is being referenced.
Claim Rejections - 35 USC § 112(d)
The following is a quotation of 35 U.S.C. 112(d):
(d) REFERENCE IN DEPENDENT FORMS.—Subject to subsection (e), a claim in dependent form shall contain a reference to a claim previously set forth and then specify a further limitation of the subject matter claimed. A claim in dependent form shall be construed to incorporate by reference all the limitations of the claim to which it refers.
Claims 127, 132, and 136 are rejected under 35 U.S.C. 112(d) as being of improper dependent form for failing to further limit the subject matter of the claim upon which it depends, or for failing to include all the limitations of the claim upon which it depends. This rejection is newly recited and necessitated by claim amendment.
Claim 127 recites “The method of claim 1, wherein the multiple design objectives comprise enzymatic activity, selectivity, stability, toxicity, size, novelty, or any combination thereof.”. Claim 1 recites “the multiple design objectives comprise activity and stability”, which already requires the multiple design objectives comprise stability as claimed in claim 127. Therefore, claim 127 fails to further limit the subject matter of claim 1.
Claims 132 and 136 are rejected under 35 U.S.C. 112(d) for the same reasons discussed above for claim 127, as applicable to independent claims 16 and 48, from which claims 132 and 136 respectively depend.
Applicant may cancel the claim(s), amend the claim(s) to place the claim(s) in proper dependent form, rewrite the claim(s) in independent form, or present a sufficient showing that the dependent claim(s) complies with the statutory requirements.
To overcome the rejection, Applicant could amend the claims to recite “wherein the multiple design objectives comprise at least two of…”, such that an additional objective other than stability is also required.
Claim Rejections - 35 USC § 101
The rejection of claims 20-21, 38, 40, and 52-53 under 35 U.S.C. 101 in the Office action mailed 04 June 2025 has been withdrawn in view of the cancellation of these claims received 03 Dec. 2025.
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 16, 18-19, 25, 46, 48, 67, and 132-139 are rejected under 35 U.S.C. 101 because the claimed invention is directed to an abstract idea without significantly more. Any newly recited portion herein is necessitated by claim amendment.
The Supreme Court has established a two-step framework for this analysis, wherein a claim does not satisfy § 101 if (1) it is “directed to” a patent-ineligible concept, i.e., a law of nature, natural phenomenon, or abstract idea, and (2), if so, the particular elements of the claim, considered “both individually and as an ordered combination,” do not add enough to “transform the nature of the claim into a patent-eligible application.” Elec. Power Grp., LLC v. Alstom S.A., 830 F.3d 1350, 1353 (Fed. Cir. 2016) (quoting Alice, 134 S. Ct. at 2355). Applicant is also directed to MPEP 2106.
Step 1: The instantly claimed invention (claims 16 and 48 being representative) is directed to a method for identifying candidate proteins. Therefore, the instantly claimed invention falls into one of the four statutory categories. [Step 1: YES]
Step 2A: First it is determined in Prong One whether a claim recites a judicial exception, and if so, then it is determined in in Prong Two if the recited judicial exception is integrated into a practical application of that exception.
Step 2A, Prong 1: Under the MPEP § 2106.04, the Step 2A (Prong 1) analysis requires determining whether a claim recites an abstract idea, law of nature, or natural phenomenon.
Claim 16 recites the following steps which fall under the mathematical concepts and/or mental processes grouping of abstract ideas:
obtaining an amino acid sequence that represents a protein design for the candidate protein;
processing the amino acid sequence by using a predictive model to generate a predicted activity of the candidate protein having the protein design that is represented by the amino acid sequence on a substrate, wherein the predictive model comprises a trained machine learning model and a physics-based simulation model and is configured to generate the predicted activity of the candidate protein by:
processing at least the amnio acid sequence by the physics-based simulation model to generate (i) a set of physicochemical attributes comprising protein-substrate relative positions and (ii) a vicinity graph that represents a fold structure of the candidate protein, wherein the physics-based model constrains the machine learning model to a physical solution space defined by the set of physicochemical attributes;
processing at least the set of physicochemical attributes by the trained machine learning model to generate the predicted activity of the candidate protein on multiple design objectives, wherein the multiple design objectives comprise activity and stability;
evaluating the candidate protein based on the predicted activity to obtain an evaluation result, wherein the evaluating comprises: selecting the candidate protein to be expressed; and
generating, based at least in part on the evaluation result, an updated amino acid sequence that represents an updated protein design for the candidate protein.
Claim 48 recites the following steps which fall under the mathematical concepts and/or mental processes grouping of abstract ideas:
generating a plurality of candidate amino acid sequences that each represent a candidate protein design of the candidate protein designs for the target protein for a substrate using a prescriptive model to process a prescriptive model input that comprises (i) an actual activity that has been measured for each of one or more historical amino acid sequences and (ii) data specifying multiple design objectives, wherein the multiple design objectives comprise activity and stability, wherein the prescriptive model comprises a multi-objective search and optimization algorithm;
for each candidate amino acid sequence that represents the candidate protein design for the target protein, processing the candidate amino acid sequence by using a predictive model to generate an output defining a predicted activity of the target protein having the candidate protein design that is represented by the candidate amino acid sequence on a substrate, wherein the predictive model comprises a machine learning model and a physics-based simulation model, wherein the physics-based simulation model is configured to process the candidate amino acid sequence to generate a set of physicochemical attributes comprising protein-substrate relative positions, wherein the physics-based model constrains the machine learning model to a physical solution space defined by the set of physicochemical attributes; and
selecting a subset of one or more candidate amino acid sequences that represent candidate protein designs for the target protein from the plurality of candidate amino acid sequences at least based on the predicted activities, wherein the one or more selected candidate amino acid sequences that are in the subset have higher levels of predicted activities than remaining, unselected candidate amino acid sequences that are not in the subset.
The identified claim limitations falls into the groups of abstract ideas of mathematical concepts and/or mental processes for the following reasons. The step of obtaining an amino acid sequence associated with a candidate protein in claim 16 can be performed mentally by analyzing a protein sequence to identify/obtain a particular amino acid sequence of interest associated with the protein. The steps of generating a plurality of candidate amino acid sequences using a prescriptive model comprising a multi-objective search and optimization algorithm in claim 48 can be performed mentally by inputting protein features, including enzyme activity and design objectives (e.g. a stability prediction) into a linear model and iteratively optimizing the model using multiple objective functions (by minimizing cost functions). Furthermore this limitation recites a mathematical concept because it requires performing mathematical optimization techniques (mathematical calculations) to obtain the candidate proteins, in light of Applicant’s specification ([00210]). The step of generating a predicted enzyme activity on a substrate for a candidate amino acid sequence using a predictive model, wherein the predictive model comprises a machine learning algorithm and a physics-based simulation as recited can be practically performed in the mind and additionally recites a mathematical concept for the following reasons. First, using a physics-based simulation model to compute the set of physicochemical attributes that comprise enzyme and substrate relative positions recites a mathematical concept because it encompasses performing mathematical calculations/simulations to compute positions. It is noted that the physics-based simulation model is not required to be carried out on a computer, such that the broadest reasonable interpretation of computing positions with a physics-based simulation model simply requires performing mathematical calculations (e.g. solving differential equations). Furthermore, using a machine learning algorithm candidate proteins as input encompasses numerical protein features (e.g. 0’s and 1’s) and the physicochemical descriptors into a trained linear regression model to output a predicted activity (i.e. by performing addition and multiplication) and then combining (e.g. averaging) the results with a predicted activity from a physics based simulation, which encompasses analyzing orientations and confirmations of proteins with a substrate using a search algorithm and scoring configurations, which is analogous to human mental work. Furthermore, the steps amount to textual equivalents to performing mathematical calculations, given each step requires performing mathematical calculations to score simulated conformations and determine a predicted activity, as discussed above. The claims further recite the mental process of selecting the candidate protein to be expressed and selecting a subset of candidate proteins, respectively, which involves evaluating the predicted activities and making a decision to select some proteins. That is, other than reciting the steps are performed by one or more computing devices, nothing in the claims precludes the steps from being practically performed in the mind. As such, the claims recite a mental process and mathematical concept.
Furthermore, the claims additionally recite the law of nature of a natural correlation between amino acid sequences/protein structures and enzymatic activity of the proteins on a substrate. See MPEP 2106.04(b), which provides examples of natural correlations a sequence or protein and the activity/function of the sequence or protein (a correlation between the presence of myeloperoxidase in a bodily sample (such as blood or plasma) and cardiovascular disease risk, Cleveland Clinic Foundation v. True Health Diagnostics, LLC, 859 F.3d 1352, 1361, 123 USPQ2d 1081, 1087 (Fed. Cir. 2017); the natural relationship between a patient’s CYP2D6 metabolizer genotype and the risk that the patient will suffer QTc prolongation after administration of a medication called iloperidone, Vanda Pharmaceuticals Inc. v. West-Ward Pharmaceuticals, 887 F.3d 1117, 1135-36, 126 USPQ2d 1266, 1281 (Fed. Cir. 2018))
Dependent claims 18, 19, 25, 46, 67, and 132-15 further recite an abstract idea and/or further limit the abstract idea of claims 16 and/or 48. Dependent claims 18-19 further recites the mental process and mathematical concept of obtaining the amino acid sequence using a prescriptive model and inputting the amino acid sequence into the predictive model comprising a multi-objective search and optimization algorithm. Dependent claim 25 further recites the mathematical concept of training a machine learning model using stochastic and deterministic optimization methods. Dependent claim 46 further recites the mental process and mathematical concept of using the machine learning model to predict a folded structure for the candidate protein based on the amino acid sequence of the candidate protein. Dependent claim 67 further limits the mathematical concept and mental process of using the prescriptive model to comprise a meta-model-assisted evolutionary algorithm. Dependent claims 132 and 136 fail to further limit claims 16 and 48 and thus are part of the abstract idea. Dependent claims 133 and 137 further limit the activity being predicted, and thus are part of the abstract idea. Dependent claims 134 and 138 further limit the selection of a candidate protein to be based on a Pareto optimality across design objectives, and thus are part of the mental process of selecting. Dependent claims 135 and 139 further limit the mathematical concept of predicting physicochemical attributes using a physics-based simulation model. Therefore, claims 16, 18-19, 25, 46, 48, 67, and 132-139 recite an abstract idea. [Step 2A, Prong 1: YES]
Step 2A: Prong 2: Under the MPEP § 2106.04, the Step 2A, Prong 2 analysis requires identifying whether there are any additional elements recited in the claim beyond the judicial exception(s), and evaluating those additional elements to determine whether they integrate the exception into a practical application of the exception. This judicial exception is not integrated into a practical application for the following reasons.
Claims 18-20, 25, 38, 40, 46, 67, and 132-139 do not recite any additional elements beyond the judicial exception, and thus are part of the judicial exception.
The additional elements of claims 16 and 48 include:
expressing the amino acid sequence to generate an expressed protein and exposing the expressed protein to the substrate (claim 16);
expressing the subset of one or more candidate amino acid sequences to generate one or more expressed proteins and exposing the one or more expressed proteins to the substrate (claim 48);
measuring an actual activity on the substrate of the expressed protein (claims 16 and 48).
The additional elements of measuring an actual activity of a selected candidate protein(s) on a substrate amounts to insignificant extra-solution activity which does not impose meaningful limits on the claim, analogous to simply cutting hair after first determining a hair-style as in In re Brown, 645 Fed. App'x 1014, 1016-1017 (Fed. Cir. 2016) or printing or downloading generated menus in Ameranth, 842 F.3d at 1241-42, 120 USPQ2d at 1854-55. See MPEP 2106.04(g). Furthermore, the additional elements do not appear to integrate the judicial exception of the predicted enzyme activity into an improvement to technology. Unlike instant claim 1, which applies the measured activity to generate another set of candidate sequences, which are then expressed to generate improved proteins relative to the prior iteration in a method of directed protein evolution, instant claims 16 and 48 simply measure activity of a candidate protein(s) to confirm the predicted activity by the judicial exception, which does not clearly improve technology or result in improved proteins.
Therefore, the additionally recited elements amount to mere instructions to apply an exception and/or amount to insignificant extra-solution activity, and, as such, the claims as a whole do no integrate the abstract idea into practical application. Thus, claims 16, 18-19, 25, 46, 67, and 132-139 are directed to an abstract idea. [Step 2A, Prong 2: NO]
Step 2B: In the second step it is determined whether the claimed subject matter includes additional elements that amount to significantly more than the judicial exception. See MPEP § 2106.05.
The claims do not include any additional steps appended to the judicial exception that are sufficient to amount to significantly more than the judicial exception for the following reasons.
Claims 18-19, 15, 46, 67, and 132-139 do not recite any additional elements beyond the judicial exception, and thus are part of the judicial exception. The additional elements of claims 16 and 48 are outlined above.
The additional element of measuring an actual activity of a protein is well-understood, routine, and conventional, as supported by Applicant’s specification and Leemhuis et al. (Directed Evolution of Enzymes: Library Screening Strategies, 2009, IUBMB Life, 61(3), pg. 222-228; previously cited) . First, Applicant’s specification at para. [00268] discloses that method for producing a colony plate are well known in the art, and include methods for synthesizing the gene of interest that encodes for the desired protein, cloning that gene behind a suitable promoter in a suitable vector, transforming the microbial host, and plating the transformed host on a suitable medium for growth of colonies (i.e. expressing selected proteins in E. coli), while Applicant’s specification at para. [00170] further discloses the host organism can be any suitable host organism, which includes E. coli. Leemhuis reviews methods for screening enzyme libraries (Abstract), and discloses that since 1990, directed evolution has been applied to screen protein libraries for desirable proteins with a desired activity (pg. 222, col. 1, para. 1 to col. 2, para. 2). Leemhius discloses activity of particular enzyme variants have been measuring S- and R-enantiomeric substrates (i.e. a plurality of substrates that have similar and dis-similar structures) (pg. 226, col. 2, para. 5). Leemhius further discloses screening proteins comprises measuring product formation (i.e. a reaction product) of enzymes to determine an activity (pg. 224, col. 2, para. 2 to pg. 225, col. 1, para. 1; Figure 2; pg. 226, col. 1, para. 2). Leemhius further discloses the above steps are repeated for multiple rounds to generate new candidate enzymes in directed evolution (Figure 1). Last, Leemhius further discloses DNA databases are screened for better performing enzymes, demonstrating the use of computers with methods of measuring protein activity are well-understood, routine, and conventional.
Therefore, taken alone, the additional elements do not amount to significantly more than the above-identified judicial exception(s). Even when viewed as a combination, the additional elements fail to transform the exception into a patent-eligible application of that exception. Thus, the claims as a whole do not amount to significantly more than the exception itself. [Step 2B: NO]
Therefore, the instantly rejected claims are not drawn to eligible subject matter as they are directed to an abstract idea (and/or natural correlation) without significantly more. For additional guidance, applicant is directed generally to applicant is directed generally to the MPEP § 2106.
Response to Arguments
Applicant's arguments filed 03 Dec. 2025 regarding 35 U.S.C. 101 have been fully considered but they are not persuasive.
Applicant remarks the claims are patent eligible because they recite additional elements that integrate the judicial exception into a practical application by reciting “expressing the amino acid sequence…” and “measuring an actual activity on the substrate of the expressed protein” in claim 16, and similarly “expressing the subset…” and “measuring an actual activity on the substrate…” in claim 48 (Applicant’s remarks at pg. 12, para. 4 to pg. 13, para. 2). Applicant remarks the dependent claims are patent eligible because they recite additional features of particular advantage as well because of their dependency (Applicant’s remarks at pg. 12, para. 7 and pg. 13, para. 3).
This argument is not persuasive. As previously discussed in the previous Office action mailed 04 June 2025 at para. [057]-[059], the limitations of expressing a selected candidate amino acid sequence and measuring activity are not sufficient to integrate the recited judicial exception into a practical application, including an improvement to technology. The additional elements only serve to simply measure activity to confirm the predicted activity by the judicial exception, which does not clearly improve technology, result in improved proteins, and/or recite a meaningful limitation beyond generally linking the judicial exception to a particular technological environment. For this reason, the additional element amounts to insignificant extra-solution activity which does not impose meaningful limits on the claim, analogous to simply cutting hair after first determining a hair-style as in In re Brown, 645 Fed. App'x 1014, 1016-1017 (Fed. Cir. 2016) or printing or downloading generated menus in Ameranth, 842 F.3d at 1241-42, 120 USPQ2d at 1854-55. See MPEP 2106.04(g).
Applicant's arguments regarding the dependent claims fail to comply with 37 CFR 1.111(b) because they amount to a general allegation that the claims define a patentable invention without specifically pointing out how the language of the claims provide the “particular advantage”. Furthermore, the arguments regarding the dependent claims are not persuasive for the same reasons discussed above for independent claims 16 and 48.
Claim Rejections - 35 USC § 102
The rejection of claims 48, 53, and 67 under 35 U.S.C. 102(a)(2) as being anticipated by Wu (2019) in the Office action mailed 04 June 2025 has been withdrawn in view of claim amendments and cancellations received 03 Dec. 2025. After further consideration, a new grounds of rejection has been set forth below under 35 U.S.C. 103.
Applicant’s arguments with respect to claims 48, 53, and 67 under 35 U.S.C. 102 have been considered but are moot because the new ground of rejection does not rely on any reference applied in the prior rejection of record for any teaching or matter specifically challenged in the argument.
Claim Rejections - 35 USC § 103
The rejections of claims 20-21, 38, and 40 under 35 U.S.C. 103 as being unpatentable over Mou (2020) in view of Wu (2019) in the Office action mailed 04 June 2025 has been withdrawn in view of the cancelation of these claims received 03 Dec. 2025.
The rejection of claims 16, 18, and 25 under 35 U.S.C. 103 as being unpatentable over Mou (2020) in view of Wu (2019) in the Office action mailed 04 June 2025 has been withdrawn in view of claim amendments received 03 Dec. 2025.
The rejection of claim 19 under 35 U.S.C. 103 as being unpatentable over Mou (2020) in view of Wu (2019), as applied to claim 18 above, and further in view of Gaeta (2021) in the Office action mailed 04 June 2025 has been withdrawn in view of claim amendments received 03 Dec. 2025.
The rejection of claim 52 under 35 U.S.C. 103 as being unpatentable over Wu (2019) in view of Mou (2021) in the Office action mailed 04 June 2025 has been withdrawn in view of the cancellation of this claim received 03 Dec. 2025.
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.
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.
Claims 1, 2, 5-9, 11-15, 48, 67, 127-131, and 136-139 are rejected under 35 U.S.C. 103 as being unpatentable over Ranganathan (2021) in view of Mou (2021) and Feala (2022), as evidenced by Alford (2017). Any newly recited portion is necessitated by claim amendment.
Cited references:
Ranganathan et al., WO 2021050923 A1 (previously cited);
Mou et al., Machine learning-based prediction of enzyme substrate scope: Application to bacterial nitrilases, 2021, Proteins, 89, pg. 336-347 and supporting information; Pub. Date: 2020 (previously cited);
Feala, US 2022/0122692 A1, effectively filed 11 Feb. 2019 based on priority to Provisional App. No. 62/804,034 (previously cited); and
Alford et al. (The Rosetta all-atom energy function for macromolecular modeling and design, 2017, 13(6), pg. 3031-3048; newly cited).
Regarding independent claim 1, Ranganathan discloses a method for evolutionary data-driven design of proteins, including enzymes (Abstract; [0073]), comprising the following steps:
Ranganathan discloses, (a) in subsequent iterations after a first iteration ([00078]), the following:
Ranganathan discloses generating a plurality of candidate amino acid sequences with desired functionality using a machine learning model in subsequent iterations (FIG. 1B, #22; FIG. 3A #102, 130; FIG. 3B #162 and #164; FIG. 5; [0078]; [0088]; [0152]), wherein the machine learning model comprises a multi-objective optimization algorithm ([0136]-[0139]; [0172] and FIG. 5) and the machine learning model takes as input design objectives including similarity and stability, and enzyme activity ([0137]; [0153]). Ranganathan further discloses measured functionality of proteins, which includes enzyme activity, are from a previous iteration and are used as input into the model to generate the candidate sequences (e.g. fitness function #65 in FIG. 1B; FIG. 3B #162-164; [0099]; [0362], e.g. activity is a measure of functionality). Ranganathan discloses each candidate amino acid sequence is a candidate enzyme design for a desired (i.e. target enzyme) ([0073]; [0101]).
Ranganathan discloses for each candidate amino acid sequence, determining values representing the desired functionality (FIG. 3A, #138; [0153], e.g. properties of candidate proteins evaluated), wherein the values include a physical energy of folding of the amino acid sequence and an activity in performing a particular functional role, the activity being predicted computationally ([0354], e.g. activity predicted computationally or experimentally), and a physics-based model for predicted protein stability (i.e. a physics-based simulation model configured to process the candidate amino acid sequence to generate a physicochemical attribute) ([0090], e.g. physics model for protein stability; [0132] and [0198], e.g. computational modeling can be used to predicted protein stability). Ranganathan discloses the activity can be predicted using a trained supervised regression model (i.e. a machine learning model configured to process the candidate amino acid sequence to generate the predicted activity) ([0198]). Ranganathan discloses the predicted protein stability is accounted for in the supervised model to increase the likelihood of candidate amino acids succeeding (i.e. the physics-based model constrains the machine learning model to a physical solution space) ([0090])
Ranganathan discloses scoring each predicted candidate sequence and selecting a subset of top candidates of the candidate sequences for experimental synthesis ([0091]; [0135], e.g. top candidates from the first set of sequences selected; [0198]; [0199], e.g. sequences lying on the frontier further refined down…). Ranganathan discloses the scoring can be based on functionality including activity ([0198]; [0354]).
Ranganathan discloses assaying the selected subset of candidate sequences by expressing the candidate amino acid sequences in bacteria to generate expressed enzymes exposed to their substrate to measure activity ([0032]; [0219], e.g. chorismite mutase catalyzes chorismite; FIG. 9)
Ranganathan discloses measuring data corresponding to enzyme activity (i.e. actual enzyme activity) exposed to its substrate [0131]-[0132], e.g. assays can measure activity; [0191]; [0240], e.g. chorismite mutase activity corresponds to enzyme activity on chorismite).
Ranganathan discloses (b), after multiple iterations ([0199]), generating an optimized protein sequence (i.e. amino acid sequence) based on a desired functionality, which includes measured activity, exceeding a predefined threshold (FIG. 3A, #165; [0065]; [0083]), which shows the optimized protein has optimal criteria in relation to the multiple design objectives.
Regarding independent claim 48, Ranganathan discloses a method for evolutionary data-driven design of proteins, including enzymes (Abstract; [0073]), comprising the following steps:
Ranganathan discloses (a) generating a plurality of candidate amino acid sequences with desired functionality using a machine learning model in subsequent iterations (FIG. 1B, #22; FIG. 3A #102, 130; FIG. 3B #162 and #164; FIG. 5; [0078]; [0088]; [0152]), wherein the machine learning model comprises a multi-objective optimization algorithm ([0136]-[0139]; [0172] and FIG. 5) and the machine learning model takes as input design objectives including similarity and stability, and enzyme activity ([0137]; [0153]). Ranganathan further discloses measured functionality of proteins, which includes enzyme activity, are from a previous iteration and are used as input into the model to generate the candidate sequences (e.g. fitness function #65 in FIG. 1B; FIG. 3B #162-164; [0099]; [0362], e.g. activity is a measure of functionality). Ranganathan discloses each candidate amino acid sequence is a candidate enzyme design for a desired (i.e. target enzyme) ([0073]; [0101]).
Ranganathan discloses (b) for each candidate amino acid sequence, determining values representing the desired functionality (FIG. 3A, #138; [0153], e.g. properties of candidate proteins evaluated), wherein the values include a physical energy of folding of the amino acid sequence and an activity in performing a particular functional role, the activity being predicted computationally ([0354], e.g. activity predicted computationally or experimentally), and a physics-based model for predicted protein stability (i.e. a physics-based simulation model configured to process the candidate amino acid sequence to generate a physicochemical attribute) ([0090], e.g. physics model for protein stability; [0132] and [0198], e.g. computational modeling can be used to predicted protein stability). Ranganathan discloses the activity can be predicted using a trained supervised regression model (i.e. a machine learning model configured to process the candidate amino acid sequence to generate the predicted activity) ([0198]). Ranganathan discloses the predicted protein stability is accounted for in the supervised model to increase the likelihood of candidate amino acids succeeding (i.e. the physics-based model constrains the machine learning model to a physical solution space) ([0090])
Regarding independent claims 1 and 48, Ranganathan does not disclose the following:
Regarding claims 1 and 48, Ranganathan does not disclose the predicted enzyme activity is on one or more substrates and the physics based simulation model is configured to generate a set of physiochemical attributes comprising protein-substrate relative positions, and wherein the machine learning model is configured to process at least the set of physiochemical attributes and the candidate amino acid sequence to generate the respective enzymatic activity. However, these limitations were known in the art, as shown by Mou and Faela.
Regarding claims 1 and 48, Mou discloses a machine-learning based method for predicting enzyme activity on substrates (Abstract), which comprises the following steps. Mou discloses obtaining candidate amino acid sequences for an enzyme (pg. 338, col. 2, para. 4 to pg. 339, col. 2, para. 2, e.g. nitrilase sequences obtained), and then processing the amino acid sequence using a predictive model to generate a predicted enzyme activity for the candidate protein (Figure 5; pg. 339, col. 2, para. 4-5). Mou discloses the predictive model comprises models to predict protein structure and a docking model (i.e. a physics-based simulation model) (pg. 339, col. 1, para. 2-3, e.g. amino acid sequences used to model each protein and interaction with each nitrile/substrate) and a machine learning model to predict the activity of each protein on the substrate (pg. 339, col.2 , para. 4-5; Figure 5). Mou discloses the amino acid sequence is processed by a structural prediction and ligand docking model (i.e. physics-based simulation model) to generate a set of physicochemical descriptors (i.e. attributes) and docking and active site descriptors, for each protein structure (corresponding to the amino acid sequence) (pg. 339, col. 1, para. 2-3 and col. 2, para. 3; Figure 5), wherein the descriptors include atomic partial charges and charges on the C and N atoms and thermodynamic properties of ligand atoms in the active site (i.e. physiochemical attributes for the candidate protein) in addition to full-atom vdW (fa_atr) attraction and vdW repulsion (fa_rep) for the interfacial interaction between the ligand and protein determined using Rosetta Ligand docking (pg. 339, col. 2, para. 2-3) (i.e. the physics based simulation model is configured to generate a set of physiochemical attributes). The vdW attraction and repulsions determined by Rosetta Ligand docking are interpreted to be relative positions between a protein and substrate, as evidenced by Alford, which discloses the repulsive van der Walls energy fa_rep and attractive van der Waals energy fa_atr are functions of atom-pair distances (distances between atoms are interpreted as a relative position between the two atoms), and a particular fa_atr value corresponds to a distance of 0 and transitions from 6-12 as the distance increases between atoms (i.e. attributes comprising relative positions between the protein and substrate) (pg. 5, para. 3 to pg. 6, para. 4; Table 1). Mou discloses the physicochemical attributes are then processed by the trained machine learning model to generate the predicted activity of the enzyme (Figure 5, e.g. descriptors used as input into machine learning model (i.e. wherein the machine learning model is configured to process at least the set of physiochemical attributes…to generate the respective enzymatic activity) (pg. 339, col. 2, para. 40; pg. 343, col. 1, para. 2, e.g. trained model applied). Mou further discloses the structure- and property-based machine learning approach to predict enzyme activity allows for the prediction of enzyme activity on particular substrates (pg. 240, col. 2, para. 2).
Ranganathan in view of Mou do not disclose the machine-learning model processes the candidate amino acid sequence in addition to the physicochemical attributes. However, Feala discloses a machine learning model predicts enzyme activity of a protein ([0039]; [0061]), based on the amino acid sequence of the protein and structural properties of the protein, including physicochemical properties (claim 24; [0009]-[0011]; [0061], e.g. amino acid sequence, atomic positions, physicochemical properties used to predict the protein functions). Faela further discloses the method provides more accurate predictions and exhibits a small memory footprint ([0059]).
It would have been prima facie obvious to one of ordinary skill in the art, before the effective filing date of the claimed invention, to have modified the model of Ranganathan to have utilized the predictive model comprising a machine learning model and physics-based simulation that generates physicochemical attributes and relative positions of claim 1, as shown by Mou, discussed above. One of ordinary skill in the art would have been motivated to combine the methods of Ranganathan and Mou in order to allow for the prediction of enzyme activity for particular substrates, as shown by Mou (pg. 339, col. 2, para. 6 to pg. 340, col. 2, para. 2). This modification would have had a reasonable expectation of success given both Ranganathan and Mou generate candidate amino acid sequences and use computational methods to determine functional properties of the sequences, such that the model of Mou is applicable to the sequences of Ranganathan.
It would have been further prima facie obvious to one of ordinary skill in the art, before the effective filing date of the claimed invention, to have modified the method of Ranganathan in view of Mou to have utilized the machine learning model that takes both the amino acid sequence and physicochemical properties as input, as shown by Faela, discussed above. One of ordinary skill in the art would have been motivated to combine the methods of Ranganathan and Mou with Faela in order to improve the accuracy of activity predictions and to use less memory, as shown by Faela ([0059]). This modification would have had a reasonable expectation of success given Mou uses a physics-based simulation model to generate physicochemical properties which could be used in the model of Faela.
Further regarding claims 1 and 48, Ranganathan does not explicitly disclose the selected subset of amino acid sequences have higher levels of respective predicted enzymatic activities than remaining, unselected candidate amino acid sequences not in the subsets.
However, as discussed above, Ranganathan discloses scoring each predicted candidate sequence and selecting a subset of the candidate sequences for experimental synthesis across the three measures of sequence similarity, stability, and functionality (i.e. activity) ([0198]; [0199], e.g. sequences lying on the frontier further refined down…), and adjustable weights can be used to prioritize certain criteria ([0199]). Ranganathan further discloses the method can be used to generate enzymes having improved activity, or any other desired biological feature exhibited by a reference protein ([0101]; [0107]), and that various design objectives may be in conflict, such as activity being inversely correlated with stability ([0137]).
It would have been prima facie obvious to one of ordinary skill in the art, before the effective filing date of the claimed invention to have selected a subset of amino acid sequences with higher levels of predicted enzyme activities through routine experimentation of the scoring of candidate sequences according to activity, stability, and functionality, within the prior art conditions of balancing various design objectives and improving enzyme activity at the expense of stability. See MPEP 2144.05 II. A.
Regarding claim 2, Ranganathan discloses the generation of candidate amino acid sequences using the prescriptive model is based on measured enzyme activity of assayed enzymes in a preceding iteration (FIG. 1B, #40, 45, 60, 65; [0130]-[0132], e.g. sequence model updated with data from previous iteration, including functional data of activity).
Regarding claim 5, Ranganathan further discloses selecting the subset of candidate amino acid sequences that have the highest uncertainty in the fitness prediction model in order to improve the prescriptive models by gathering data to reduce the uncertainty ([0140]-[0141]; [0201]).
Regarding claim 6, Ranganathan discloses expressing the enzymes in E. Coli ([0219]; [0237]; FIG. 1E, e.g. expression in microbes).
Regarding claim 7, Ranganathan discloses a quantity of E. coli is cultured ([0219]). Regarding the process in which the E. coli that is cultured was previously measured, this limitation is a product by process limitation that only serves to define the process in which the E. coli was previously measured. See MPEP 2113 I. Here, since the product of Ranganathan (i.e. the culture of E. coli) is the same as the produce in the claims (E. coli), Ranganathan discloses claim 7.
Regarding claim 8, Ranganathan discloses the enzymes can be genes expressed in host strains and then harvested for protein purification and purified, which shows the cells were lysed ([0116]-[0118]).
Regarding claim 9, Ranganathan discloses the protein purification can involve affinity based purification, which enriches the protein ([0118]).
Regarding claims 11-15¸ Ranganathan does not disclose: measuring the actual enzymatic activity comprises measuring a concentration of the one or more substrates and/or a reaction product, as recited in claim 11; wherein the one or more substrates comprises a plurality of substrates, as recited I claim 12; wherein at least one of the plurality of substrates is a desired substrate that is specific to a desired physical production of the target enzyme, as recited in claim 13; wherein at least two of the plurality of substrates have a same structure, as recited in claim 14, and wherein at least two of the plurality of substrates have different structures. However, Mou discloses these limitations as follows:
Regarding claims 11-15, Mou discloses a machine-learning based method for predicting enzyme activity on substrates (Abstract), which comprises measuring enzyme activity of enzymes for various substrates (Figure 5). Regarding claim 11, Mou further discloses the enzyme activity is measured by measuring the ammonia produced by the reaction between the enzyme and substrate (i.e. a reaction product) (pg. 338, col. 2, para. 3). Regarding claim 12, Mou further discloses an activity for each enzyme is predicted and measured for a plurality of substrates (Figure 3, e.g. substrates along top of microplate; Figure 5). Regarding claim 13, Mou further discloses a substrate is a desired substrate (Figure 3; pg. 338, col. 2, para. 3, e.g. "desired substrate" is added). Regarding claim 14, Mou further discloses the substrates include two aromatic substrates (i.e. substrates with a same structure) (Figure 3). Regarding claim 15, Mou further discloses the plurality of substrates includes an aliphatic and aromatic substrate (i.e. substrates with dissimilar structures) (Figure 3).
Further regarding claims 11-15, Mou discloses the measured enzyme activity for the different types of substrates can be used to train a machine learning model to predict enzyme activity for different substrates, thus determining substrate scopes of an enzyme (figure 5; pg. 343, col. 1, para. 2 to col. 2, para. 2).
It would have been prima facie obvious to one of ordinary skill in the art, before the effective filing date of the claimed invention, to have modified the method of Ranganathan to have measured enzyme activity by measuring a reaction product for a plurality of different substrates, including a desired substate, substrates with the same structure, and substrates with different structures, as shown by Mou above. One of ordinary skill in the art would have been motivated to combine the methods of Ranganathan further with Mou in order to allow for the prediction of enzyme activity of an enzyme for different substrates, as shown by Mou (figure 5; pg. 343, col. 1, para. 2 to col. 2, para. 2). This modification would have had a reasonable expectation of success because Ranganathan also uses a machine learning model to predict enzyme activity as a desired function ([0089]), such that the data of Mou is applicable to the model of Ranganathan.
Regarding claim 67, Ranganathan discloses the prescriptive model includes a successive Pareto optimization (SPO) (i.e. a meta-model-assisted evolutionary algorithm) ([0138]).
Regarding claims 127 and 136, Ranganathan discloses the multi-objective optimization algorithm ([0136]-[0139]; [0172] and FIG. 5) and the machine learning model takes as input design objectives including similarity and stability, and enzyme activity ([0137]; [0153]), as discussed above for claim 1.
Regarding claims 128 and 137¸ Ranganathan discloses the activity of the protein may be a novel function and the method may be used to identify novel proteins with desired activity ([0069]; [0101]; [0155]).
Regarding claims 129 and 138, Ranganathan discloses the best amino acid sequences are identified using a Pareto frontier to solve the multi-objective optimization problem ([0137]-[0138]).
Regarding claims 130¸ Ranganathan discloses obtaining information regarding the functionality of proteins including an enzyme activity (i.e. an operating condition) from a previous iteration (e.g. fitness function #65 in FIG. 1B; FIG. 3B #162-164; [0099]; [0362], e.g. activity is a measure of functionality), which is used as input into the model with multiple design objectives including enzyme activity and stability to produce candidate sequences (i.e. at the operating condition)([0137]; [0153]).
Regarding claims 131 and 139, Ranganathan in view of Mou disclose the physicochemical attributes include van der Waals attraction (i.e. binding energies) in addition to structure dynamics, as applied to claim 1 above.
Therefore, the invention is prima facie obvious.
Claims 16, 18-19, 25, and 132-135 are rejected under 35 U.S.C. 103 as being unpatentable over Ranganathan (2021) in view of Mou (2021), as evidenced by Alford (2017). This rejection is newly recited and necessitated by claim amendment.
Cited references:
Ranganathan et al., WO 2021050923 A1 (previously cited);
Mou et al., Machine learning-based prediction of enzyme substrate scope: Application to bacterial nitrilases, 2021, Proteins, 89, pg. 336-347 and supporting information; Pub. Date: 2020 (previously cited); and
Alford et al. (The Rosetta all-atom energy function for macromolecular modeling and design, 2017, 13(6), pg. 3031-3048; newly cited).
Regarding independent claim 16, Ranganathan discloses a method for evolutionary data-driven design of proteins, including enzymes (Abstract; [0073]), comprising the following steps:
Ranganathan discloses (a) generating (i.e. obtaining) a candidate amino acid sequence with desired functionality using a machine learning model (FIG. 1B, #22; FIG. 3A #102, 130; FIG. 3B #162 and #164; FIG. 5; [0078]; [0088]; [0152]), wherein the machine learning model comprises a multi-objective optimization algorithm ([0136]-[0139]; [0172] and FIG. 5) and the machine learning model takes as input design objectives including similarity and stability, and enzyme activity ([0137]; [0153]). Ranganathan discloses each candidate amino acid sequence is a candidate enzyme design for a desired (i.e. target enzyme) ([0073]; [0101]).
Ranganathan discloses (b) for the candidate amino acid sequence, determining values representing the desired functionality (FIG. 3A, #138; [0153], e.g. properties of candidate proteins evaluated), wherein the values include a physical energy of folding of the amino acid sequence and an activity in performing a particular functional role, the activity being predicted computationally ([0354], e.g. activity predicted computationally or experimentally), and a physics-based model for predicted protein stability (i.e. a physics-based simulation model configured to process the candidate amino acid sequence to generate a physicochemical attribute) ([0090], e.g. physics model for protein stability; [0132] and [0198], e.g. computational modeling can be used to predicted protein stability). Ranganathan discloses the activity can be predicted using a trained supervised regression model (i.e. a machine learning model configured to process the candidate amino acid sequence to generate the predicted activity) ([0198]). Ranganathan discloses the predicted protein stability is accounted for in the supervised model to increase the likelihood of candidate amino acids succeeding (i.e. the physics-based model constrains the machine learning model to a physical solution space) ([0090])
Ranganathan discloses (c) scoring (i.e. evaluating) each predicted candidate sequence and selecting a subset of top candidates of the candidate sequences for experimental synthesis ([0091]; [0135], e.g. top candidates from the first set of sequences selected; [0198]; [0199], e.g. sequences lying on the frontier further refined down…). Ranganathan discloses the scoring can be based on functionality including activity (i.e. evaluating based on the predicted activity) ([0198]; [0354]).
Ranganathan discloses selecting the candidate protein to be expressed based (i.e. the selected subset) ([0091]; [0135], e.g. top candidates from the first set of sequences selected).
Ranganathan discloses assaying the selected subset of candidate sequences by expressing the candidate amino acid sequences in bacteria to generate expressed enzymes exposed to their substrate to measure activity ([0032]; [0219], e.g. chorismite mutase catalyzes chorismite; FIG. 9)
Ranganathan discloses measuring data corresponding to enzyme activity (i.e. actual enzyme activity) exposed to its substrate [0131]-[0132], e.g. assays can measure activity; [0191]; [0240], e.g. chorismite mutase activity corresponds to enzyme activity on chorismite).
Ranganathan discloses (d), after multiple iterations of the above process ([0199]), generating an optimized protein sequence (i.e. amino acid sequence) based on a desired functionality, which includes measured activity, exceeding a predefined threshold (i.e. based on the evaluation result) (FIG. 3A, #165; [0065]; [0083]), which shows the optimized protein has optimal criteria in relation to the multiple design objectives.
Regarding claim 16, Ranganathan does not disclose the following:
Regarding claim 16, Ranganathan does not disclose the predicted enzyme activity is on one or more substrates and that the physics based simulation model is configured to generate a set of physiochemical attributes comprising protein-substrate relative positions and (ii) a vicinity graph that represents a folded structure of the candidate protein. However, these limitations were known in the art, as shown by Mou.
Regarding claims 1 and 48, Mou discloses a machine-learning based method for predicting enzyme activity on substrates (Abstract), which comprises the following steps. Mou discloses obtaining candidate amino acid sequences for an enzyme (pg. 338, col. 2, para. 4 to pg. 339, col. 2, para. 2, e.g. nitrilase sequences obtained), and then processing the amino acid sequence using a predictive model to generate a predicted enzyme activity for the candidate protein (Figure 5; pg. 339, col. 2, para. 4-5). Mou discloses the predictive model comprises models to predict protein structure and a docking model (i.e. a physics-based simulation model) (pg. 339, col. 1, para. 2-3, e.g. amino acid sequences used to model each protein and interaction with each nitrile/substrate) and a machine learning model to predict the activity of each protein on the substrate (pg. 339, col.2 , para. 4-5; Figure 5). Mou discloses the amino acid sequence is processed by a structural prediction and ligand docking model (i.e. physics-based simulation model) to generate a set of physicochemical descriptors (i.e. attributes) and docking and active site descriptors, for each protein structure (corresponding to the amino acid sequence) (pg. 339, col. 1, para. 2-3 and col. 2, para. 3; Figure 5), wherein the descriptors include atomic partial charges and charges on the C and N atoms and thermodynamic properties of ligand atoms in the active site (i.e. physiochemical attributes for the candidate protein) in addition to full-atom vdW (fa_atr) attraction and vdW repulsion (fa_rep) for the interfacial interaction between the ligand and protein determined using Rosetta Ligand docking (pg. 339, col. 2, para. 2-3) (i.e. the physics based simulation model is configured to generate a set of physiochemical attributes and protein-substrate relative positions). The vdW attraction and repulsions determined by Rosetta Ligand docking are interpreted to be relative positions between a protein and substrate, as evidenced by Alford, which discloses the repulsive van der Walls energy fa_rep and attractive van der Waals energy fa_atr are strictly functions of atom-pair distances (distances between atoms are interpreted as a relative position between the two atoms), and a particular fa_atr value corresponds to a distance of 0 and transitions from 6-12 as the distance increases between atoms (i.e. attributes comprising relative positions between the protein and substrate) (pg. 5, para. 3 to pg. 6, para. 4; Table 1). Mou discloses the structural prediction and ligand docking model (i.e. the physics based simulation model) also generates a three-dimensional graph structure of an enzyme depicting the vicinity between residues (i.e. a vicinity graph representing a folded structure of the protein) (pg. 339, col. 1, para. 2). Mou further discloses the physicochemical attributes are then processed by the trained machine learning model to generate the predicted activity of the enzyme (Figure 5, e.g. descriptors used as input into machine learning model (i.e. wherein the machine learning model is configured to process at least the set of physiochemical attributes…to generate the respective enzymatic activity) (pg. 339, col. 2, para. 40; pg. 343, col. 1, para. 2, e.g. trained model applied). Mou further discloses the structure- and property-based machine learning approach to predict enzyme activity allows for the prediction of enzyme activity on particular substrates (pg. 240, col. 2, para. 2).
Ranganathan in view of Mou do not disclose the machine-learning model processes the candidate amino acid sequence in addition to the physicochemical attributes. However, Feala discloses a machine learning model predicts enzyme activity of a protein ([0039]; [0061]), based on the amino acid sequence of the protein and structural properties of the protein, including physicochemical properties (claim 24; [0009]-[0011]; [0061], e.g. amino acid sequence, atomic positions, physicochemical properties used to predict the protein functions). Faela further discloses the method provides more accurate predictions and exhibits a small memory footprint ([0059]).
It would have been prima facie obvious to one of ordinary skill in the art, before the effective filing date of the claimed invention, to have modified the model of Ranganathan to have utilized the predictive model comprising a machine learning model and physics-based simulation that generates physicochemical attributes and relative positions of claim 1, as shown by Mou, discussed above. One of ordinary skill in the art would have been motivated to combine the methods of Ranganathan and Mou in order to allow for the prediction of enzyme activity for particular substrates, as shown by Mou (pg. 339, col. 2, para. 6 to pg. 340, col. 2, para. 2). This modification would have had a reasonable expectation of success given both Ranganathan and Mou generate candidate amino acid sequences and use computational methods to determine functional properties of the sequences, such that the model of Mou is applicable to the sequences of Ranganathan.
Regarding the dependent claims:
Regarding claims 18-19, Ranganathan discloses generating the plurality of candidate amino acid sequences with desired functionality using a machine learning model in subsequent iterations (FIG. 1B, #22; FIG. 3A #102, 130; FIG. 3B #162 and #164; FIG. 5; [0078]; [0088]; [0152]), wherein the machine learning model comprises a multi-objective optimization algorithm ([0136]-[0139]; [0172] and FIG. 5) and the machine learning model takes as input design objectives including similarity and stability, and enzyme activity ([0137]; [0153]). Ranganathan further discloses measured functionality of proteins, which includes enzyme activity, are from a previous iteration and are used as input into the model to generate the candidate sequences (e.g. fitness function #65 in FIG. 1B; FIG. 3B #162-164; [0099]; [0362], e.g. activity is a measure of functionality). Ranganathan discloses each candidate amino acid sequence is a candidate enzyme design for a desired (i.e. target enzyme) ([0073]; [0101]).
Regarding claim 25, Ranganathan discloses the machine learning model can be trained using one or more of gradient descent (i.e. a deterministic optimization method) and stochastic gradient descent (i.e. stochastic optimization method) ([0277]).
Regarding claims 132, Ranganathan discloses the multi-objective optimization algorithm ([0136]-[0139]; [0172] and FIG. 5) and the machine learning model takes as input design objectives including similarity and stability, and enzyme activity ([0137]; [0153]), as discussed above for claim 1.
Regarding claims 133¸ Ranganathan discloses the activity of the protein may be a novel function and the method may be used to identify novel proteins with desired activity ([0069]; [0101]; [0155]).
Regarding claims 134, Ranganathan discloses the best amino acid sequences are identified using a Pareto frontier to solve the multi-objective optimization problem ([0137]-[0138]).
Regarding claims 135, Ranganathan in view of Mou disclose the physicochemical attributes include van der Waals attraction (i.e. binding energies) in addition to structure dynamics, as applied to claim 16 above.
Claim 46 is rejected under 35 U.S.C. 103 as being unpatentable over Ranganathan in view of Mou, as applied to claim 16 above, and further in view of Feala (2019). This rejection is newly recited and necessitated by claim amendment.
Cited reference: Feala, US 2022/0122692 A1, effectively filed 11 Feb. 2019 based on priority to Provisional App. No. 62/804,034 (previously cited); and
Regarding claim 46¸ Ranganathan in view of Mou disclose the method of claim 16 as applied above.
Ranganathan in view of Mou do not disclose the machine learning model predicts the folded structure of the candidate protein based at least in part on the amino acid sequence of the candidate protein”.
However, Feala discloses a machine learning model that predicts the free energy of the folded structure of a protein (i.e. a folded structure) in addition to enzyme activity of a protein ([0039]), based on the amino acid sequence of the protein and structural properties of the protein (claim 24; [0009]-[0011], e.g. amino acid sequence, atomic positions, physicochemical properties used to predict the protein functions). Feala further discloses the prediction of the folded structure of a protein is a metric of protein stability ([0039]), which is a desired protein property ([0003]).
It would have been prima facie obvious to one of ordinary skill in the art, before the effective filing date of the claimed invention, to have modified the method of Ranganathan in view of Mou, as applied to claim 16 above, to have further predicted a folded structure of the candidate protein in addition to enzyme activity, as shown by Feala (claim 24; [0009]-[0010]; [0039]). One of ordinary skill in the art would have been motivated to combine the methods of Ranganathan in view of Mou and Feala to achieve a desired protein stability in the candidate protein, as shown by Feala ([0003]; claim 27). This modification would have had a reasonable expectation of success given both Mou and Faela predict enzyme activity using machine learning models to predict enzyme activity that take physicochemical properties as input, and thus the model of Faela is applicable to Rangathan in view of Mou.
Therefore, the invention is prima facie obvious.
Response to Arguments
Applicant's arguments filed 03 Dec. 2025 regarding 35 U.S.C. 103 have been fully considered but they are not persuasive.
Applicant remarks that Ranganathan, Mou, and Faela do not disclose “wherein the multiple design objectives comprise activity and stability, and wherein the prescriptive model comprises a multi-objective search and optimization algorithm” and “wherein the predictive model comprises a machine learning model and a physics-based simulation model, wherein the physics-based simulation model is configured to process the candidate amino acid sequence to generate a set of physicochemical attributes comprising protein-substrate relative positions, wherein the physics-based model constrains the machine learning model to a physical solution space…” as in amended claim 1 (Applicant’s remarks at pg. 14, para. 5 to pg. 15, para. 1). Applicant remarks that Ranganathan merely describes a method and apparatus for designing sequence-defined biomolecules, and that Mou merely describes combining targeted experimental activity data with structural modeling, ligand docking, and physicochemical properties with machine learning models, but Mou does not teach “the physics-based model constrains the machine learning model to a physical solution space” and “wherein the machine learning model is configured to process at least the set of physicochemical attributes…” (Applicant’s remarks at pg. 15, para. 2). Applicant remarks that Feala merely describes identifying associations between amino acid sequences and protein functions or properties, and therefore the combination of Ranganathan, Mou, and Faela does not disclose each and every element of claim 1 (Applicant’s remarks at pg. 15, para. 3-4).
This argument is not persuasive. Applicant generally summarizes the overall theme of Ranganathan, but does not address the specifically recited portions and teachings discussed in the previous and above rejection. Ranganathan does not “merely describe” a method and apparatus for designing sequenced-defined biomolecules. Ranganathan does disclose a prescriptive model of a machine learning model comprising a multi-objective optimization algorithm ([0136]-[0139]; [0172] and FIG. 5) and the machine learning model takes as input design objectives including similarity and stability, and enzyme activity ([0137]; [0153]). Furthermore, as explained above, Ranganathan discloses the predicted protein stability is accounted for in the supervised model to increase the likelihood of candidate amino acids succeeding (i.e. the physics-based model constrains the machine learning model to a physical solution space) ([0090]).
Applicant’s remarks regarding Mou are also not persuasive. First, Mou is not relied upon for the multi-objective search algorithm or the prescriptive model. Ranganathan discloses a predictive model with a machine learning model and physics-based simulation model that generates a physicochemical attribute as discussed in the above rejection. Mou is relied upon to disclose making the predictions of Ranganathan on a particular substrate(s) and a physics-based simulation model that produces the physicochemical attributes, including the protein-substrate relative positions that are then processed by a machine learning model to make a prediction of enzyme activity.
Applicant states that Mou using a physics-based simulation model to produce structural features to train the machine learning model does not show that the physics-based model constrains the machine learning model to a physical solution space. However, the instant claims also simply recite that the physics-based model produces “physicochemical features” that are then processed by a separate machine learning model. As understood by one of ordinary skill in the art, training a machine learning model specifically on physicochemical features generated from a physics-based model does constrain the machine learning model to a physical solution space defined by the physicochemical attributes used in the training process. If Applicant intends for some other interaction between the physics-based model and the machine learning model to provide the claimed constraint to the machine learning model, then this should be recited in the claims.
With respect to the newly recited feature of the “protein-substrate relative positions”, Mou further discloses the amino acid sequence is processed by a structural prediction and ligand docking model (i.e. physics-based simulation model) to generate a set of physicochemical descriptors (i.e. attributes) and docking and active site descriptors, for each protein structure (corresponding to the amino acid sequence) (pg. 339, col. 1, para. 2-3 and col. 2, para. 3; Figure 5), wherein the descriptors include full-atom vdW (fa_atr) attraction and vdW repulsion (fa_rep) for the interfacial interaction between the ligand and protein determined using Rosetta Ligand docking (pg. 339, col. 2, para. 2-3) (i.e. the physics based simulation model is configured to generate a set of physiochemical attributes). The vdW attraction and repulsions determined by Rosetta Ligand docking are interpreted to be relative positions between a protein and substrate, as evidenced by Alford (newly recited above), which discloses the repulsive van der Walls energy fa_rep and attractive van der Waals energy fa_atr are functions of atom-pair distances (distances between atoms are interpreted as a relative position between the two atoms), and a particular fa_atr value corresponds to a distance of 0 and transitions from 6-12 as the distance increases between atoms (i.e. attributes comprising relative positions between the protein and substrate) (pg. 5, para. 3 to pg. 6, para. 4; Table 1). Mou discloses the physicochemical attributes are then processed by the trained machine learning model to generate the predicted activity of the enzyme (Figure 5, e.g. descriptors used as input into machine learning model (pg. 339, col. 2, para. 40; pg. 343, col. 1, para. 2, e.g. trained model applied).
Faela is used to disclose the combination of processing both the physicochemical descriptors in addition to the machine learning model to make a prediction of enzyme activity on a substrate, as discussed in the above rejection, and is not relied upon for the above limitations.
Applicant remarks that the claimed subject matter provides a technical advantage and unexpected results over the cited references, namely an improved method of engineering and cites various discussions of advantages in the specification (Applicant’s remarks at pg. 15, para. 5 to pg. 16, para. 3).
This argument is not persuasive. In response to applicant's argument that the claimed subject matter provides technical advantage over the prior art, the fact that the inventor has recognized another advantage which would flow naturally from following the suggestion of the prior art cannot be the basis for patentability when the differences would otherwise be obvious. See Ex parte Obiaya, 227 USPQ 58, 60 (Bd. Pat. App. & Inter. 1985).
Therefore, while Applicant points out various advantages of the claimed method discussed in the specification, these same advantages would flow naturally from the combination of Ranganathan in view of Mou and Faela.
Applicant remarks the combination of Ranganathan, Mou, and Feala fail to provide one of ordinary skill in the art with a reasonable expectation of success (Applicant’s remarks at pg. 16, para. 4).
This argument is not persuasive because Applicant's arguments fail to comply with 37 CFR 1.111(b) because they amount to a general allegation that the claims define a patentable invention without specifically pointing out how the language of the claims patentably distinguishes them from the references.
Applicant remarks the dependent claims 2, 5-9, and 11-15 are also not obvious over Ranganathan, Mou, and Feala for the reasons discussed above (Applicant’s remarks at pg. 16, para. 5-6).
This argument is not persuasive for the same reasons discussed above for claim 1.
Applicant remarks, regarding claim 16, that Mou and Wu fail to disclose “processing…by the physics-based simulation model to generate (i) a set of physicochemical attributes comprising protein-substrate relative positions and (ii) a folded structure of the candidate protein…(including the full limitation”, and Mou merely describes “[c]combining targeted experimental activity data with structural modeling, ligand docking, and physicochemical properties of proteins and ligands with various machine learning models” and Wu merely describes “demonstrating that machine learning-guided directed evolution finds variants with higher fitness…”, and therefore the combination of Mou and Wu does not disclose each and every element of claim 16 (Applicant’s remarks at pg. 16, para. 8 to pg. 17, para. 3).
This argument is not persuasive. First, Wu is not relied upon in the above rejection. Furthermore, Ranganathan in view of Mou discloses the claimed limitations for the reasons discussed in the above rejection, and as also explained above for claim 1. Applicant does not actually address the particular teachings of Mou recited in the previous rejection, and instead just provides an overly broad summary of the teachings of Mou.
Applicant remarks the combination of Mou and Wu ail to provide one of ordinary skill in the art with a reasonable expectation of success (Applicant’s remarks at pg. 17, para. 4).
This argument is not persuasive because Applicant's arguments fail to comply with 37 CFR 1.111(b) because they amount to a general allegation that the claims define a patentable invention without specifically pointing out how the language of the claims patentably distinguishes them from the references.
Applicant remarks dependent claims 18, 25, 38, and 40 are also not obvious over Mou and Wu (Applicant’s remarks at pg. 17, para. 6).
This argument is not persuasive for the same reasons discussed above for claim 16.
Applicant remarks that claim 19 is free of the prior art for the reasons discussed for claim 16, and because Gaeta does not cure the deficiencies of Mou and Wu (Applicant’s remarks at pg. 18, para. 1 to pg. 19, para. 1).
This argument is not persuasive for the reasons discussed above for claim 16 and because Gaeta is not relied upon for claim 19.
Applicant remarks that claim 46 is free of the prior art for the same reasons as claim 16 and because Feala does not cure the deficiencies of Mou and Wu (Applicant’s remarks at pg. 19, para. 2 to pg. 20, para. 1).
This argument is not persuasive for the same reasons discussed above for claim 16.
Conclusion
No claims are allowed.
Claims 1-2, 5-9, 11-15, and 127-131 are patent eligible for the reasons discussed in the Office action mailed 04 June 2025.
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 KAITLYN L MINCHELLA whose telephone number is (571)272-6485. The examiner can normally be reached 7:00 - 4:00 M-Th.
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, Olivia Wise can be reached at (571) 272-2249. 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.
/KAITLYN L MINCHELLA/Primary Examiner, Art Unit 1685