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 .
Information Disclosure Statement
The IDS filed 8/18/2023 has been considered by the Examiner.
Priority
Applicant’s claim for the benefit of a prior-filed application under 35 U.S.C. 119(e) or under 35 U.S.C. 120, 121, or 365(c) is acknowledged. Priority of US application 63/068,687 filed 08/21/2020 is acknowledged.
Claim Status
Claims 4, 6, 12-16, 19-21, 23-24, 27,29, 31-53, 56, 60, 62 and 70-74 are cancelled.
Claims 1-3, 5, 7-11, 17-18, 22, 25-26, 28, 30, 54-55, 57-59, 61, 63-69 and 75 are pending and are examined on the merits.
Claim Objections
Claim 26 recites “and visual trait measured as the sub-organismal level” at the ending phrase, which should read as “and a visual trait measured as the sub-organismal level”.
Appropriate correction is required.
Claim Rejections - 35 USC § 112—First Paragraph
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 67 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(s) contains subject matter which was not described in the specification in such a way as to reasonably convey to one skilled in the relevant art that the inventor or a joint inventor, or for applications subject to pre-AIA 35 U.S.C. 112, the inventor(s), at the time the application was filed, had possession of the claimed invention.
Claim 67 recites “an organism genetically improved at a target”. However, the disclosure does not describe a genetically improved organism. The independent claim 54 (where the claim 67 depends from) is to a pure in silico method (with a trained machine learning model). Claims 57-59 recite the conventional breeding and genome editing technology, but that are not enough to evidence possession of all possible organism encompassed by claim 67, which would also include a human. The disclosure at paragraphs [190-195] describes an experiment of compensatory genetic variants (related to two human genes BBS4 and RPGRIP1L, involved in ciliopathies) in zebrafish. However, an organism that is genetically improved is not provided.
Claim Rejections - 35 USC § 112—Second Paragraph
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 68-69 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.
The term “more easily accessible” in claim 68 (lines 1-2) is a relative term which renders the claim indefinite. The term “more easily accessible” is not defined by the claim, 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. A person of ordinary skills in art hence does not know to identify genetic variants for a transgenic technology or a genome editing technology.
Claim 69 is indefinite when reciting “producing the genetic variant”. It is not clear how the genetic variant is produced. The disclosure provides no further description.
Claim Rejections - 35 USC § 112
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.
The following is a quotation of pre-AIA 35 U.S.C. 112, fourth paragraph:
Subject to the following paragraph [i.e., the fifth paragraph of pre-AIA 35 U.S.C. 112], 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 57 and 58 are rejected under 35 U.S.C. 112(d) or pre-AIA 35 U.S.C. 112, 4th paragraph, 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. Both claims 57 and 58 recite “the method of claim 54, wherein the genetic improvement is achieved by…” However, “the genetic improvement” is recited in claim 55, not claim 54. Hence claims 57 and 58 should both depend on claim 55.
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.
Claim Rejections - 35 USC § 101
35 U.S.C. 101 reads as follows:
Whoever invents or discovers any new and useful process, machine, manufacture, or composition of matter, or any new and useful improvement thereof, may obtain a patent therefor, subject to the conditions and requirements of this title.
Claims 1-3, 5, 7-11, 17-18, 22, 25-26, 28, 30, 54-55, 57-59, 61, 63-69 and 75 are rejected under 35 U.S.C. 101 because the claimed invention is directed to non-statutory subject matter.
Step 1: Process, Machine, Manufacture or Composition
Claims 1-3, 5, 7-11, 17-18, 22, 25-26, 28, 30 and 75 are to a method for assessing effects of genetic variants with a series functional steps, so a process.
Claims 54-55, 57-59, 61 and 63-67 are to a method for identifying targets for genetically improving a trait in an organism with a series functional steps, so a process.
Claims 68-69 are to a method for identifying genetic variants as alternative candidates for use as targets that are more easily accessible with a series functional steps, so a process.
Step 2A Prong One: Identification of Abstract Ideas or Product of Nature
The claims recite:
b) Automatically inputting the dataset of sequences to a trained machine-learning model to predict one or more effect scores (claims 1, 54 and 68).
Wherein the model is configured to output one or more effect scores corresponding to the probabilities of one or more secondary genetic variants having a compensatory effect to the primary genetic variant, or the magnitudes of said effects (claims 1, 54 and 68);
Wherein the effect affects an attribute associated with a trait of the organism (claim 54).
---Under a broadest reasonable interpretation (BRI), the “trained machine-learning model” can be a mathematical algorithm such as a linear regression model. Step b) hence equates to a mathematical operation that generates an output of “effect scores”, using input of “dataset of sequences”. Step b) is directed to an abstract idea of mathematical concepts.
The two “wherein” phrases further describe the mathematical model regarding the output data, which equates to data and information that can be achieved in human mind. Hence the step also recite an abstract idea of mental processes.
Dependent claims further describe model outputs and model input data, model training and loss functions, which are classified into abstract ideas. Particularly, dependent claim 67 recites “an organism genetically improved”, which is drawn to a product of nature. “An organism genetically improved” also encompass a human. A product of nature is part of the judicial exceptions.
Step 2A Prong Two: Consideration of Practical Application
The claims result in a process of displaying the predicted effect scores on a display device, which is classified into an insignificant extra-solution activity of outputting. The claims do not recite any additional elements that integrate the abstract idea/judicial exception into a practical application.
This judicial exception is not integrated into a practical application because the claims do not meet any of the following criteria:
An additional element reflects an improvement in the functioning of a computer, or an improvement to other technology or technical field;
an additional element that applies or uses a judicial exception to effect a particular treatment or prophylaxis for a disease or medical condition;
an additional element implements a judicial exception with, or uses a judicial exception in conjunction with, a particular machine or manufacture that is integral to the claim;
an additional element effects a transformation or reduction of a particular article to a different state or thing; and
an additional element applies or uses the judicial exception in some other meaningful way 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.
Step 2B: Consideration of Additional Elements and Significantly More
The claimed method also recites "additional elements" that are not limitations drawn to an abstract idea. The recited additional elements are drawn to:
Receiving a dataset of sequences from an input device (claims 1, 54 and 68).
Displaying the predicted effect scores on a display device (claims 1, 54 and 68).
Receiving a pre-training dataset comprising a plurality of batches of naturally
occurring sequences (claim 2).
Inputting each batch of sequences into a language model (claim 2).
A organism genetically improved at a target identified by the method of claim 54 (claim 67).
The claim(s) does/do not include additional elements that are sufficient to amount to significantly more than the judicial exception because these are all insignificant extra-solution activities of data inputting/outputting (MPEP §2106.05(g)).
The claims do not include additional elements that are sufficient to amount of significantly more than the judicial exception because it is routine and conventional to perform the acts of acquire input data for analyzing and output analytical results in a display device. Other elements of the method include “A organism genetically improved” which is a recitation of generic breeding that is a well-understood, routine, and conventional activity previously known to the pertinent industry. Viewed as a whole, these additional claim element(s) do not provide meaningful limitation(s) to transform the abstract idea recited in the instantly presented claims into a patent eligible application of the abstract idea such that the claim(s) amounts to significantly more than the abstract idea itself. Therefore, the claim(s) are rejected under 35 U.S.C. 101 as being directed to non-statutory subject matter.
Claim Rejections - 35 USC § 103
In the event the determination of the status of the application as subject to AIA 35 U.S.C. 102 and 103 (or as subject to pre-AIA 35 U.S.C. 102 and 103) is incorrect, any correction of the statutory basis (i.e., changing from AIA to pre-AIA ) for the rejection will not be considered a new ground of rejection if the prior art relied upon, and the rationale supporting the rejection, would be the same under either status.
The following is a quotation of 35 U.S.C. 103 which forms the basis for all obviousness rejections set forth in this Office action:
A patent for a claimed invention may not be obtained, notwithstanding that the claimed invention is not identically disclosed as set forth in section 102, if the differences between the claimed invention and the prior art are such that the claimed invention as a whole would have been obvious before the effective filing date of the claimed invention to a person having ordinary skill in the art to which the claimed invention pertains. Patentability shall not be negated by the manner in which the invention was made.
The factual inquiries for establishing a background for determining obviousness under 35 U.S.C. 103 are summarized as follows:
1. Determining the scope and contents of the prior art.
2. Ascertaining the differences between the prior art and the claims at issue.
3. Resolving the level of ordinary skill in the pertinent art.
4. Considering objective evidence present in the application indicating obviousness or nonobviousness.
This application currently names joint inventors. In considering patentability of the claims the examiner presumes that the subject matter of the various claims was commonly owned as of the effective filing date of the claimed invention(s) absent any evidence to the contrary. Applicant is advised of the obligation under 37 CFR 1.56 to point out the inventor and effective filing dates of each claim that was not commonly owned as of the effective filing date of the later invention in order for the examiner to consider the applicability of 35 U.S.C. 102(b)(2)(C) for any potential 35 U.S.C. 102(a)(2) prior art against the later invention.
Claims 1, 11, 17-18, 22, and 25-26 are rejected under 35 U.S.C. 103 as being unpatentable over Riesselman et. al.: (“Deep generative models of genetic variation capture the effects of mutations,” Nature Methods 15, 816-822 (2018), Newly cited), and in view of Hopf et al. (“Mutation effects predicted from sequence co-variation,” Nature Biotechnology 35, 128-135 (2017). Newly cited).
Claim 1 is interpreted as a method for assessing effects of genetic variants. Regarding claim 1, Riesselman provides (page 816, section Abstract) “predicted the effects of mutations across a variety of deep mutational scanning experiments”, which teaches (preamble) computational prediction/assessment of mutation effects.
Riesselman provides (page 816, section Abstract) “a probabilistic model for sequence families, predicted the effects of mutations across a variety of deep mutational scanning experiments substantially better than existing methods based on the same evolutionary data. The model, learned in an unsupervised manner solely on the basis of sequence information, probabilistic model for sequence families”, and (page 817, col 2, Fig. 2 legend) “After fitting a probabilistic model to a family of homologous sequences, we heuristically quantified the effect of mutation as the log-ratio of mutant likelihood to wild-type likelihood”; Riesselman does not teach secondary generic variants. Hopf provides (page 128, title) “sequence co-variation” and (page 128, section Abstract) “residue dependencies between positions”, which suggests secondary genetic variants. Together Riesselman and Hopf teaches (step a)) sequence-family data, wild-type/mutant comparison, and multi-position dependencies..
Riesselman provides (page 816, section Abstract) “a probabilistic model for sequence families, predicted the effects of mutations across a variety of deep mutational scanning experiments substantially better than existing methods based on the same evolutionary data. The model, learned in an unsupervised manner solely on the basis of sequence information, probabilistic model for sequence families”, which teaches (step b)) trained generative model inputting biological sequences.
Riesselman does not teach attribute effects associated with trait. Hopf provides (page 128, section Abstract) “explicitly captures residue dependencies between positions. … outperforms methods that do not account for epistasis”, which teaches attribute effects associated with trait. Riesselman provides (page 816, col 2, last line through page 817, col 1, first 2 lines) “we can use the probabilities that the model assigns to any given sequence as a proxy for the relative plausibility of a molecule satisfying functional constraints. We consider the log-ratio … as a heuristic metric for the relative favorability of a mutated sequence”, which teaches probability/log-ratio scoring of mutant sequence plausibility. A person of ordinary skills in art would interpret a secondary mutation restoring plausibility/fitness relative to a deleterious primary mutation as a compensatory/positive epistatic effect.
Riesselman provides (page 818, Fig 3), which teaches (step c)) displaying or otherwise outputting predicted effect scores in a display device.
Regarding claim 11, Riesselman provides (page 816, section Abstract) “a major unanswered question in biological research, clinical medicine, and biotechnology is how to decipher and exploit the effects of mutations on biomolecules”, which teaches mutation effects relevant to clinical medicine and biotechnology. The claim’s variant categories are broad and known.
Regarding claim 17, Riesselman provides (page 819, col 1, last para) “the effects of mutations across diverse classes of proteins and RNAs”, which teaches the dataset sequences are DNA, RNA.
Regarding claim 18, Riesselman provides (page 816, col 2, 2nd para) “we developed nonlinear latent-variable models for biological sequence families and leveraged approximate inference techniques to infer the families from large multiple-sequence alignments. We show how a Bayesian deep latent-variable model can be used to reveal latent structure in sequence families and predict the effects of mutations with accuracies exceeding those of site-independent or pairwise-interaction models”, which suggests Single gene/protein, gene/protein family, interacting proteins/complex. Interacting protein complex is less directly mapped but ordinary extension of protein-function prediction.
Regarding claim 22, Riesselman provides (page 818, col 1, 1st para) “The clear exceptions to the overall superiority of DeepSequence were the comparisons to the viral protein mutation experiment effects”, which teaches the dataset of viral sequences.
Regarding claim 25, Riesselman provides (page 818, col 1, 1st para) “A mutation may be damaging with regard to some measurable protein feature-for example, enzyme efficiency-but harmless for stability or even organism fitness”, which teaches the effect at molecular, sub-organismal or organismal level.
Regarding claim 26. Riesselman does not teach endophenotypes. Hopf provides (page 132, Figure 4C) “Modification methylase HaeIII” as “improvements of the epistatic model for functional sites”. Hopf hence teaches protein modifications as an endophenotype effect.
It would have been prima facie obvious to a person of ordinary skills in art to combine Riesselman’s trained sequence-family mutation-effect model and Hopf’s epistatic/co-variation mutation-effect framework to score secondary variants in the context of a primary of a primary variant; because predicting biological sequence-function or variant-effect relationships from large sequence datasets when labeled experimental data are scarce and combinatorial variant testing is infeasible, while secondary variant having an compensatory effect to a primary variant is a predictable positive/context-dependent epistatic interaction of the type Hopf expressly teaches to model.
One would reasonably expect success because both references address the same recognized problem, and the Riesselman/Hopf combination would have improved variant-effect scoring, including scoring of secondary variants in the context of a primary variant, context-dependent constraints, residue dependencies, and epistasis.
Claims 2, 5, 28, and 75 are rejected under 35 U.S.C. 103 as being unpatentable over Riesselman and Hopf, as applied to Claims 1, 11, 17-18, 22, and 25-26 above, and further in view of Alley et. al.:(“Unified rational protein engineering with sequence-based deep representation learning,” Nature Methods 16, 1315-1322 (2019). Newly cited).
Regarding claim 2, neither Riesselman nor Hopf teach transfer learning. Alley provides (page 1316, Fig. 1 legend) “UniRep model was trained on 24 million UniRef50 primary amino-acid sequences”, which teaches naturally occurring protein databases are used for pretraining. Batches are conventional in neural-network training, and Alley training necessarily uses iterative/batched optimization.
Alley provides (page 1315, section abstract) “statistical representation that is semantically rich”, and (page 1316, Fig. 1 legend) “fixed-length vector representation”; Alley provides (page 1316, col 1, 2nd para) “amino-acid character embeddings” and (page 1316, col 1, 3rd para) “amino-acid embeddings”; Alley teaches protein representation and transfer learning (page 1316, Fig. 1). Alley expressly characterizes the representation as semantically rich and numerical; Hence Alley teaches inputting each batch into language model configured to output semantic features.
Alley provides (page 1316, Fig. 1 legend) “The model was trained to perform next amino-acid prediction (minimizing cross-entropy loss)”, which suggests automatically updating language model after each batch. Updating model weights during training is inherent/express in loss-minimization training.
Alley provides (page 1316, Fig. 1 legend) “A top model (for example, a sparse linear regression or random forest) trained on top of the representation, which acts as a featurization of the input sequence, enables supervised learning on diverse protein informatics tasks”; and (page 1322, col 2, 6th para) “We built various other concatenations of these representations (Supplementary Table 3), denoted as ‘Fusions’, to be evaluated in supervised stability and quantitative function prediction”
Alley teaches the pre-trained representation followed by supervised downstream models, which is a transfer learning with labeled sequences mapped to effects.
Alley provides (page 1315, col 2, 3rd para) “we use a recurrent neural network (RNN) to learn statistical representations of proteins from ~24 million UniRef50 sequences (Fig. 1a)”, which teaches training neural network model based on final training dataset, because the UniRef50 primary amino-acid sequences is the final training dataset.
Regarding claim 5, neither Riesselman nor Hopf teaches a binary loss function. Alley provides (page 1316, Fig. 1 legend) “The model was trained to perform next amino-acid prediction (minimizing cross-entropy loss)”, Alley teaches the binary cross-entropy loss function.
Regarding claim 28, neither Riesselman nor Hopf teaches the enzymatic efficiency caused by genetic variation. Alley provides (page 130, col 1, 2nd para) “the epistatic model also successfully captured the effects of muta-tions in low-throughput experiments on thermostability of trypsin (ρ = 0.77, N = 23) and SH3 (ρ = 0.69, N = 48) and the catalytic efficiency of β-lactamase15 (ρ = 0.87, N = 30), including double and triple substitutions”, which teaches enzymatic efficiency.
Regarding claim 75, neither Riesselman nor Hopf teaches a transfer learning task. Alley provides (page 1319, col 2, last para) “start with the globally trained UniRep and then fine-tune it to the evolutionary context of the task (Fig. 4a)”, which teaches a fine-tuning task following pre-training, then supervised top model (page 1316, Fig. 1).
Alley provides (page 1320, col 1, 1st para) “~25,000 likely evolutionarily related sequences obtained via JackHMMER search”, which teaches mapping to common ancestry/homology/MSA-related sequences.
Alley provides (page 1316, col 1, 2nd para) “amino-acid character embeddings” and (page 1316, col 1, 3rd para) “amino-acid embeddings”; Alley teaches protein representation and transfer learning (page 1316, Fig. 1). Alley expressly characterizes the representation as semantically rich and numerical; Hence Alley teaches inputting each batch into language model configured to output semantic features.
Alley provides (page 1316, Figure legend) “A top model (for example, a sparse linear regression or random forest) trained on top of the representation, which acts as a featurization of the input sequence, enables supervised learning on diverse protein informatics tasks”; and Riesselman provides (page 817, Fig. 1 legend) “mutation-effect scores for positions 310–393 in the PDZ domain”. Alley teach the feature-transfer architecture; Riesselman supply mutation-effect scoring/epistasis motivation.
It would have been prima facie obvious to a person of ordinary skills in art to combine Riesselman/Hopf’s mutation-effect and epistasis-scoring models with Alley’s protein language-model pretraining/transfer-learning architectures because the combined pipeline of Riesselman and Hopf that predict-mutation effects, context-dependent constraints, residue dependencies, and epistasis rely on labelled sequences, while Alley’s teaching of training useful sequence representations and downstream supervised models from naturally occurring unlabeled sequences.
One would reasonably expect success as the combination would have yielded predictable results because all of Riesselman, Hopf and Alley address the same recognized problem: predicting biological sequence-function or variant-effect relationships from large sequence datasets when labeled experimental data are scarce and combinatorial variant testing is infeasible, and Alley’s unlabeled sequences are powerful compensation to Riesselman/Hopf’s labelled sequences.
Claims 3, 7-10, and 30 are rejected under 35 U.S.C. 103 as being unpatentable over Riesselman and Hopf, as applied to Claims 1, 11, 17-18, 22, and 25-26 above, and further in view of Wu et. al.: (“Machine learning-assisted directed protein evolution with combinatorial libraries,” Proceedings of the National Academy of Sciences of the United States of America 116(18):8852-8858 (2019). Newly cited).
Regarding claim 3, Riesselman does not teach variant having effects on metric of interest. Hopf provides (page 128, section Abstract) “many high-throughput experimental technologies have been developed to assess the effects of large numbers of mutations (variation) on phenotypes”, which teaches that prior art uses sequence-function/fitness data to evaluate variant effects. Hopf hence suggests receiving training dataset with reference sequence and primary variant having an effect on metric of interest.
Riesselman provides (page 816, col 2, last para) “we approximate the evolutionary process as a ‘sequence generator’ that generates a sequence x with probability p(x|θ) and parameters θ that are fit to reproduce the statistics of evolutionary data”, which teaches generative sequence modeling.
Neither Riesselman nor Hopf teaches multi-mutant libraries (with secondary genetical variants), Wu provides (page 8852, section Abstract) “we incorporate machine learning into the directed evolution workflow. Combinatorial sequence space can be quite expensive to sample experimentally, but machine-learning models trained on tested variants provide a fast method for testing sequence space computationally”, and (page 8852, col 2, 2nd para) “we first investigate the benefits of in silico screening by machine learning using the dataset collected”, which teaches computational exploration of multi-mutant libraries. The specific phrase “random seed” is an implementation detail of stochastic generation.
Wu provides (page 8852, section Abstract) “machine-learning models trained on tested variants provide a fast method for testing sequence space computationally. We validated this approach on a large published empirical fitness landscape for human GB1 binding protein, demonstrating that machine learning-guided directed evolution finds variants with higher fitness than those found by other directed evolution approaches”, which teaches model training on experimentally measured variant fitness.
Wu provides (page 8852, section Abstract) “machine learning-guided directed evolution finds variants with higher fitness than those found by other directed evolution approaches. We then provide an example application in evolving an enzyme to produce
each of the two possible product enantiomers (i.e., stereo divergence) of a new-to-nature carbene Si–H insertion reaction. The approach predicted libraries enriched in functional enzymes and fixed seven mutations in two rounds of evolution to identify variants for selective catalysis with 93% and 79% ee (enantiomeric excess)”. Wu teaches selecting variants/libraries based on model predictions. Acceptance/rejection based on predicted fitness or loss would have been routine.
Wu provides (page 8852, section Abstract) “two rounds of evolution to identify variants for selective catalysis with 93% and 79% ee (enantiomeric excess)”. Wu teaches Iterative model-based design and directed-evolution rounds. Exact convergence-to-minimum language is a standard optimization stopping condition.
Regarding claim 7, Riesselman provides (page 817, col 2, Fig. 2 legend) “we heuristically quantified the effect of mutation as the log-ratio of mutant likelihood to wild-type likelihood (as approximated by the ELBO; Methods). Bottom: mutation-effect scores for positions 310–393 in the PDZ domain”, which teaches effect score for genetic variants. Neither Riesselman nor Hopf teaches secondary genetic variants. Wu provides (page 8852, section Abstract) “predicted libraries enriched in functional enzymes”, which suggests secondary genetic variants. Once variants are scored, selecting higher-scoring variants is a predictable use.
Regarding claim 8, neither Riesselman nor Hopf teaches prioritizing genetic variants. Wu provides (page 8856, col 1, last para) “By sampling large regions of sequence space in silico to reduce in vitro screening efforts, we rapidly evolved a single parent enzyme to generate variants that selectively form both product enantiomers of a new-to-nature C–Si bond-forming reaction”, which teaches prioritization by ML-guided screening..
Regarding claim 9, neither Riesselman nor Hopf teaches evaluates epistasis modeling. Wu provides (page 8853, col 1, Fig. 1 legend) “Machine learning-assisted directed evolution. As a result of increased throughput provided by screening in silico, four positions can be explored simultaneously in a single round, enabling a broader search of sequence–function relationships and deeper exploration of epistatic interactions”, which evaluates epistasis modeling.
Regarding claim 10, neither Riesselman nor Hopf teaches altering genetic variants in an organism. Wu provides (page 8852, section Abstract) “To reduce experimental effort associated with directed protein evolution and to explore the sequence space encoded by mutating multiple positions simultaneously, we incorporate machine learning into the directed evolution workflow”;
Wu provides (page 8852, section Abstract) “Combinatorial sequence space can be quite expensive to sample experimentally, but machine-learning models trained on tested variants provide a fast method for testing sequence space computationally”;
Wu provides (page 8852, section Abstract) “The approach predicted libraries enriched in functional enzymes and fixed seven mutations in two rounds of evolution to identify variants for selective catalysis with 93% and 79% ee (enantiomeric excess)”.
Wu teaches wet-lab variant construction/testing and model-guided iterative rounds. Wu’s enzyme directed-evolution examples, anticipate endophenotype limitations because updating ML from tested variant phenotypes was known.
Regarding claim 30, neither Riesselman nor Hopf teaches effects affecting fitness of an organism. Wu provides (page 8852, col 1, section Abstract) “We validated this approach on a large published empirical fitness landscape for human GB1 binding protein, demonstrating that machine learning-guided directed evolution finds variants with higher fitness than those found by other directed evolution approaches”, which teaches fitness of organisms, and fitness landscape.
It would have been prima facie obvious to a person of ordinary skills in art to combine Riesselman/Hopf’s mutation-effect and epistasis-scoring models with Wu’s teaching that modeling machine learning training on experimentally measured variant fitness, and selecting variants/libraries based on model predictions. Because combinatorial sequence space is expensive to sample experimentally and ML models trained on tested variants can rapidly screen candidate sequence space in silico.
One would reasonably expect success because the combination would merely apply known ML-assisted protein/variant-engineering feedback loops to the known problem of identifying beneficial or compensatory variants.
Claims 54-55, 57-59, 61, 63-64 and 66-69 are rejected under 35 U.S.C. 103 as being unpatentable over Hopf et al. (“Mutation effects predicted from sequence co-variation,” Nature Biotechnology 35, 128-135 (2017). Newly cited), in view of Riesselman et. al.: (“Deep generative models of genetic variation capture the effects of mutations,” Nature Methods 15, 816-822 (2018), Newly cited), and Jaganathan et al.: (“CRISPR for Crop Improvement: An Update Review,” Frontiers in Plant Science 9:985 (2018). Newly cited).
Claim 54 is interpreted as a method to identify targets for genetically improving a trait in an organism. Regarding claim 54, Hopf provides (page 128, section Abstract”) “robust methods to predict the effects of genetic variation are needed. … EVmutation can be used to assess the quantitative effects of mutations in genes of any organism”; Hopf supplies mutation-effect prediction across organisms;
Neither Riesselman nor Hopf teach genetically improving a trait in an organism. Jaganathan provides (page 3, Figure 1), which teaches genetically improving a trait in an organism.
Riesselman provides (page 816, section Abstract) “We found that DeepSequence, a probabilistic model for sequence families, predicted the effects of mutations across a variety of deep mutational scanning experiments substantially better than existing methods based on the same evolutionary data. The model, learned in an unsupervised manner solely on the basis of sequence information, is grounded with biologically motivated priors, reveals the latent organization of sequence families”; and (page 816, col 1, 2nd para) “Most improvements to computational predictions of mutation effects have been driven by leveraging of the signal of evolutionary conservation among homologous sequences”. Riesselman teaches (step a)) receiving sequence datasets in computational sequence-model workflows.
Riesselman provides (page 817, col 2, Fig. 2 legend) “we heuristically quantified the effect of mutation as the log-ratio of mutant likelihood to wild-type likelihood”; and Hopf provides (page 128, section Abstract) “Most prediction methods exploit evolutionary sequence conservation but do not consider the interdependencies of residues or bases. We present EVmutation, an unsupervised statistical method for predicting the effects of mutations that explicitly captures residue dependencies between positions”; and (page 129, col 2, 2nd para) “we quantified the effects of single or higher-order substitutions on a particular sequence background with the log-odds ratio of sequence prob-abilities between the wild-type and mutant sequences”. Riesselman teaches the reference/wild-type sequence and mutant sequences are expressly used. Hopf’s higher-order substitutions and epistasis support combinations where the effect of a secondary variant is assessed in context of a primary variant.
Riesselman provides (page 816, section Abstract) “We found that DeepSequence, a probabilistic model for sequence families, predicted the effects of mutations across a variety of deep mutational scanning experiments substantially better than existing methods based on the same evolutionary data. The model, learned in an unsupervised manner solely on the basis of sequence information, is grounded with biologically motivated priors, reveals the latent organization of sequence families”, which teaches sequence families associated with an organism, because “same evolutionary data” suggests same organism.
Riesselman provides (page 816, section Abstract) “predicted the effects of mutations”; and (page 817, col 2, Fig. 2 legend) “log-ratio of mutant likelihood to wild-type likelihood.” The claimed effect score reads on likelihood/log-likelihood/fitness-related mutation-effect scores.
Hopf provides (page 128, section Abstract”) “explicitly captures residue dependencies between positions”; and “outperforms methods that do not account for epistasis”; Hopf teaches (step b)) context-dependent/epistatic mutation effects; a compensatory secondary variant is a favorable epistatic interaction that offsets a primary variant.
Hopf provides (page 128, section Abstract”) “effects of genetic variation … on phenotypes”, which teaches the effect affects an attribute associated with a trait of the organism.
Riesselman provides (page 818, Fig 3), which teaches (step c)) displaying or otherwise outputting predicted effect scores in a display device.
Jaganathan provides (page 3, Figure 1), which suggests displaying genetically improving a trait in an organism in an output device.
Regarding claim 55, neither Riesselman nor Hopf teaches genetical improvement of an organism. Jaganathan provides (page 3, Figure 1), which teaches selecting one or more identified targets for genetical improvement of an organism.
Regarding claim 57, neither Riesselman nor Hopf teaches conventional breeding. Jaganathan provides (page 13, col 1, 1st para lines 1-3) “Grape is an economically valuable fruit, with breeders targeting numerous fruit quality traits such as aroma, disease and abiotic stress resistance, fruit size and skin color”, which teaches conventional breeding.
Regarding claim 58, neither Riesselman nor Hopf teaches the genome-editing technology. Jaganathan provides (page 1, section Abstract) “advances in genome editing approaches has opened up possibilities to breed for almost any given desirable trait”, which teaches the genome-editing technology.
Regarding claim 59, neither Riesselman nor Hopf teaches the genome-editing technology. Jaganathan teaches (page 1, section Abstract) the genome editing technology is CRISPR, TALEN, or ZFN.
Regarding claim 61, neither Riesselman nor Hopf teaches the genome-editing technology. Jaganathan provides (page 11, col 1, 1st para) “using multiplex CRISPR /Cas9 was made possible through Golden Gate cloning and Multisite Gateway LR recombination methods”, which teaches the genome editing coupled with recombination system.
Regarding claim 63, neither Riesselman nor Hopf teaches the genome-editing technology. Jaganathan provides Figure 4 (page 8), which teaches the CRISPR/Cas genome-editing technology applies to maize, rice, and more plants.
Regarding claim 64, neither Riesselman nor Hopf teaches the genome-editing technology. Jaganathan provides (page 4, col 2, 2nd para) “CRISPR/Cas9-based genome editing has been utilized to increase crop disease resistance and also to improve tolerance to major abiotic stresses like drought and salinity”, which teaches the drought tolerance trait for genome editing.
Regarding claim 66, neither Riesselman nor Hopf teaches a trait affecting growth rate. Jaganathan provides (page 13, col 1, 2nd para) “Targeting NCED4 in a co-editing strategy could, therefore, be used to enrich for germline-edited events by merely germinating seeds at high temperature. Germination thermotolerance due to inactivation of NCED4 provides a useful whole-plant selectable phenotype of pleiotropic effects on growth or stress tolerance”, which teaches a trait affecting growth rate.
Regarding claim 67, neither Riesselman nor Hopf teaches an organism genetically improved. Jaganathan provides (page 13, col 1, 1st para) “Wang Y. et al. (2016) have identified five types of CRISPR/Cas9 target sites in the widely cultivated grape species Vitis vinifera for potential genome editing. Editing using purified CRISPR/Cas9 ribonucleoproteins (RNPs) as delivery particles in grape protoplasts has been shown to be effective against the powdery mildew susceptibility gene, MLO-7. Targeted mutagenesis of VvWRKY52, a transcription factor gene has elucidated its role in biotic stress responses. In addition, knockout of VvWRKY52 in grape increased disease resistance to fungal infection (Botrytis cinerea)”, which teaches an organism genetically improved at a target identified.
Regarding claim 68 preamble: Hopf provides (page 129, col 2, 2nd para) “we estimated the site and coupling parameters h and J using regularized maximum pseudo--likelihood. After the parameters were inferred, we quantified the effects of single or higher-order substitutions on a particular sequence background with the log-odds ratio of sequence prob-abilities between the wild-type and mutant sequences (Fig. 1 and Supplementary Fig. 1)”; Jaganathan teaches the genome editing at user-defined loci and CRISPR (page 1, section Abstract). Combined Hopf and Jaganathan teach (preamble) identifying genetic variants as alternative candidates for more accessible transgenic/genome-editing targets
Riesselman provides (page 816, section Abstract) “a probabilistic model for sequence families, predicted the effects of mutations across a variety of deep mutational scanning experiments substantially better than existing methods based on the same evolutionary data. The model, learned in an unsupervised manner solely on the basis of sequence information, probabilistic model for sequence families”, and (page 817, col 2, Fig. 2 legend) “After fitting a probabilistic model to a family of homologous sequences, we heuristically quantified the effect of mutation as the log-ratio of mutant likelihood to wild-type likelihood”; Hopf provides (page 128, title) “sequence co-variation” and (page 128, section Abstract) “residue dependencies between positions”, which teaches epistatic residue dependencies. Together Riesselman and Hopf teaches (step a)) sequence-family data, wild-type/mutant comparison, and multi-position dependencies.
Riesselman provides (page 817, col 2, Fig. 2 legend) “we heuristically quantified the effect of mutation as the log-ratio of mutant likelihood to wild-type likelihood”; and Hopf provides (page 128, section Abstract) “Most prediction methods exploit evolutionary sequence conservation but do not consider the interdependencies of residues or bases. We present EVmutation, an unsupervised statistical method for predicting the effects of mutations that explicitly captures residue dependencies between positions”; and (page 129, col 2, 2nd para) “we quantified the effects of single or higher-order substitutions on a particular sequence background with the log-odds ratio of sequence prob-abilities between the wild-type and mutant sequences”. Riesselman teaches the reference/wild-type sequence and mutant sequences are expressly used. Hopf’s higher-order substitutions and epistasis support combinations where the effect of a secondary variant is assessed in context of a primary variant.
Riesselman provides (page 816, section Abstract) “a probabilistic model for sequence families, predicted the effects of mutations across a variety of deep mutational scanning experiments substantially better than existing methods based on the same evolutionary data. The model, learned in an unsupervised manner solely on the basis of sequence information, probabilistic model for sequence families”, which teaches trained generative model inputting biological sequences.
Hopf provides (page 128, section Abstract) “explicitly captures residue dependencies between positions. … outperforms methods that do not account for epistasis”, which teaches epistatic residue dependencies. Riesselman provides (page 816, col 2, last line through page 817, col 1, first 2 lines) “we can use the probabilities that the model assigns to any given sequence as a proxy for the relative plausibility of a molecule satisfying functional constraints. We consider the log-ratio … as a heuristic metric for the relative favorability of a mutated sequence”, which teaches probability/log-ratio scoring of mutant sequence plausibility. A POSITA would interpret a secondary mutation restoring plausibility/fitness relative to a deleterious primary mutation as a compensatory/positive epistatic effect.
Riesselman provides (page 818, Fig 3), which teaches (step c)) displaying or otherwise outputting predicted effect scores in a display device. However, neither Riesselman nor Hopf teaches genetically improving a trait in an organism. Jaganathan provides (page 3, Figure 1), which suggests displaying genetically improving a trait in an organism in an output device.
Regarding claim 69, neither Riesselman nor Hopf teaches stress tolerance. Jaganathan teaches (page 3, Figure 1) genome editing of single and multiplex targets, and to produce selected variants after evaluation for abiotic stress tolerance.
It would have been prima facie obvious to a person of ordinary skills in art to combine Riesselman’s mutation-effect and Hopf’s epistasis-scoring models because the references address the same recognized problem: predicting biological sequence-function or variant-effect relationships from large sequence datasets when labeled experimental data are scarce and combinatorial variant testing is infeasible. Riesselman/Hopf teach what to predict-mutation effects, context-dependent constraints, residue dependencies, and epistasis.
One would reasonably expect success because the combination would have yielded predictable results: improved variant-effect scoring, including scoring of secondary variants in the context of a primary variant.
It would have been prima facie obvious to a person of ordinary skills in art to combine Riesselman/Hopf’s mutation-effect and epistasis-scoring models with Jaganathan’s CRISPR/Cas9-based genome editing technology. Because Jaganathan’s crop genome sequence and genome-editing approaches opened possibilities “to breed for almost any given desirable trait” (Jagannathan: page 1, Section Abstract).
One would reasonably expect success as the combination would have yielded predictable results, because Riesselman/Hopf’s in silico method will be enhanced by Jaganathan’s experimental CRISPR/Cas9 techniques that develop crops with desirable traits contributing to increased yield potential under biotic and abiotic stress conditions.
Claim 65 is rejected under 35 U.S.C. 103 as being unpatentable over Hopf, Riesselman and Jaganathan, as applied to claims 54-55, 57-59, 61, 63-64 and 66-67 above, and further in view of Zhang et al. ("Comparison of gene editing efficiencies of CRISPR/Cas9 and TALEN for generation of MSTN knock-out cashmere goats." Theriogenology 132 (2019): 1-11. Newly cited).
Regarding claim 65, none of Hopf, Riesselman and Jaganathan teaches an organism selected from an animal. Zhang teaches the organism is selected from “cashmere goat” (page 2, col 1, penultimate para line 1).
It would have been prima facie obvious to a person of ordinary skills in art to combine Riesselman/Hopf/Jaganathan’s pipeline which predicts mutation-effect and epistasis-scores followed by CRISPR/Cas9-based genome editing technology to generate trait modified organism, with Zhang’s teaching to select an animal. Because knock out the MSTN gene in Zhang’s cashmere goat will bring economic gain (Zhang: page 1, col 1, last para).
One would reasonably expect success as the combination would have yielded predictable results, because this combination combines the trait selection with specific organism target for genetic modification, which does not interfere each other. This would be a classic example of:
Combining prior art elements according to known methods to yield predictable results (MPEP §2141.III.(A))
Conclusion
No claims are allowed.
Any inquiry concerning this communication or earlier communications from the examiner should be directed to GUOZHEN LIU whose telephone number is (571)272-0224. The examiner can normally be reached Monday-Friday 8-5.
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 D 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.
/GL/
Patent Examiner
Art Unit 1686
/Anna Skibinsky/
Primary Examiner, AU 1635
Prosecution Insights