DETAILED ACTION
Notice of Pre-AIA or AIA Status
The present application, filed on or after March 16, 2013, is being examined under the first inventor to file provisions of the AIA .
Claim Status
Claims 1-15, 25-27, 34, and 36 are pending and examined herein.
Claims 1-15, 25-27, 34, and 36 are rejected.
Claim 25 is objected to.
Priority
Claims 1-15, 25-27, 34, and 36 are granted the claim to the benefit of priority to U.S. Provisional application 63/107362 filed 29 October 2020. Thus, the effective filling date of claims 1-15, 25-27, 34, and 36 is 29 October 2020.
Information Disclosure Statement
The information disclosure statements (IDS) were received 23 October 2024, 20 February 2025, and 24 October 2025. The submission is in compliance with the provisions of 37 CFR 1.97. Accordingly, the information disclosure statements have been considered by the examiner.
Drawings
The drawings received 10 March 2023 are accepted.
Specification
The disclosure is objected to because it contains an embedded hyperlink and/or other form of browser-executable code in the last line of [0037]. Applicant is required to delete the embedded hyperlink and/or other form of browser-executable code; references to websites should be limited to the top-level domain name without any prefix such as http:// or other browser-executable code. See MPEP § 608.01.
Claim Objections
Claim 25 is objected to because of the following informalities:
Claim 25 recites “The method of claim 1, in which wherein the structure parameters comprise…” in lines 1-2 of the claim but should read “The method of claim 1, wherein the structure parameters comprise…”.
Appropriate correction is required.
Claim Interpretation
Claim 1, 34, and 36 recite “full multiple sequence alignment”. The BRI of this limitation is a multiple sequence alignment (MSA) assigned to a training example (amino acid sequence of a particular protein) that has not been processed through data augmentation methods of removing or masking amino acid sequences which reduces the assigned MSA of a training example. The instant disclosure refers to the full MSA as the unreduced MSA (see instant disclosure [0053]).
Claim Rejections - 35 USC § 112
112/b
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 27, 34, and 36 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.
Claim 27 recites “extracting the protein from a human or animal body and obtaining the amino acid sequence from the extracted protein” which renders the metes and bounds of the claim indefinite. The indefiniteness arises because it is unclear which protein in the claims that “the protein” in this limitation is referring to. It is unclear if this protein is referring to the “protein” in line 2 of claim 26, if this protein referring to the “protein” in line 3 of claim 1, or if this protein referring to one of the “plurality of proteins” in claim 1 (if this protein is referring to one of the plurality of proteins in claim 1, then it is further unclear which protein of the plurality of proteins that “the protein” is referring to). For the sake of furthering examination, this limitation will be interpreted as referring to the protein in which the amino acid sequence is obtained in claim 26.
Claims 34 and 36 recite the limitation “the method comprising…” (line 9 of claim 34) and (line 6 of claim 36). There is insufficient antecedent basis for this limitation in the claim. The indefiniteness arises because the claim does not make clear what “the method” is referring to in the claims. This rejection could be overcome by amendment of this limitation to “the operations comprising”. For the sake of furthering examination, this limitation will be interpreted as “the operations comprising”.
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 1-15, 25-27, 34, and 36 are rejected under 35 U.S.C. 101 because the claimed invention is directed to a judicial exception (i.e., a law of nature, a natural phenomenon, or an abstract idea) without significantly more.
(Step 1)
Claims 1-15 and 25-27 fall under the statutory category of a process and claims 34 and 36 fall under the statutory category of a machine.
(Step 2A Prong 1)
Under the BRI, the instant claims recite judicial exceptions that are an abstract idea of the type that is in the grouping of a “mental process”, such as procedures for evaluating, analyzing or organizing information, and forming judgement or an opinion. The instant claims further recite judicial exceptions that are an abstract idea of the type that is in the grouping of a “mathematical concept”, such as mathematical relationships and mathematical equations.
Independent claims 1, 34, and 36 recite mental processes of “obtaining, for each of a plurality of proteins, a full multiple sequence alignment for the protein” and “determining, for each of the plurality of proteins, a reduced multiple sequence alignment for the protein, comprising removing or masking data from the full multiple sequence alignment for the protein”.
Independent claims 1, 34, and 36 recite mathematical concepts of “generating, for each of the plurality of proteins, target structure parameters… comprising processing a representation of the full multiple sequence alignment for the protein to generate output structure parameters…”, “determining the target structure parameters for the protein based on the output structure parameters for the protein”, “training the structure prediction neural network to, for one or more of the plurality of proteins, process a representation of the reduced multiple sequence alignment for the protein to generate structure parameters that match the target structure parameters for the protein”.
Dependent claim 7 recites a mathematical concept of “training the structure prediction neural network to, for each of the plurality of proteins, process the representation of the reduced multiple sequence alignment for the protein to generate an auxiliary output that predicts the identity of each masked amino acid in the reduced multiple sequence alignment for the protein”. Dependent claim 10 recites a mathematical concept of “for each of the plurality of proteins, determining a confidence estimate for the target structure parameters for the protein”. Dependent claim 11 recites mental processes of “identifying one or more proteins for which the confidence estimate for target structure parameters for the protein does not satisfy a threshold” and “refraining from training the structure prediction neural network on the identified proteins”. Dependent claim 14 recites a mathematical concept of “training the structure prediction neural network to, for one or more other proteins, process a representation of a multiple sequence alignment for the other protein to generate structure parameters that match ground truth structure parameters for the other protein”. Dependent claim 26 recites a mathematical concept of “determine a structure of the protein”.
The claims recite mental processes of obtaining, for a plurality of proteins, a full multiple sequence alignment for a protein (which encompasses comparing amino acid sequences of proteins to determine similarity and organizing amino acid sequences to align the similar amino acid sequences of proteins), determining, for each of the plurality of proteins, a reduced multiple sequence alignment for the protein, comprising removing or masking data from the full multiple sequence alignment for the protein (which encompasses analyzing data and organizing the multiple sequence alignment data by removing amino acid sequences, sampling a reduction parameter value from a distribution to specify a number of sequences to remove, removing the specified number of sequences, randomly selecting sequences to be removed, masking an identity of an amino acid at one or more positions in the sequences, randomly sampling the positions to be masked see dependent claims 2-6), identifying proteins which have a confidence estimate for target structure parameters that do not satisfy a threshold (which encompasses an observation and judgment of a confidence estimate against a threshold to determine the estimate does not satisfy a threshold), refraining from training the structure prediction neural network on the identified proteins with confidence estimates that do not satisfy a threshold (which encompasses organizing training data by removing these identified proteins from a training dataset).
The claims recite mathematical concepts as mathematical calculations of processing a representation of the full multiple sequence alignment for the protein to generate output structure parameters characterizing a structure of the protein (which encompasses processing numeric values, such as statistical features derived from the multiple sequence alignment, to calculate the numerical values of atomic coordinates and torsion angles see instant disclosure [0039] and [0040], processing numerical values to generate numerical values encompasses performing a mathematical calculation), determining the target structure parameters for the protein based on the output structure parameters for the protein (which encompasses adding random noise values which are numerical values to the numerical values that represent the output structure parameters of the protein which is a mathematical operation adding numerical values see [0060] and dependent claim 8), training the structure prediction neural network to, for one or more of the plurality of proteins, process a representation of the reduced multiple sequence alignment for the protein to generate structure parameters that match the target structure parameters for the protein (which encompasses mathematical calculations of determining gradients of an objective function that measures an error between generated structure parameters and target structure parameters and repeatedly adjusting current values of the model parameters to determine trained values of the model parameters from the initial values of the model parameters using stochastic gradient descent and backpropagation which is a series of mathematical calculations see instant disclosure [0034], [0049], [0065], and dependent claim 12), training the structure neural network to process the representation of the reduced multiple sequence alignment to generate an auxiliary output that predicts the identity of each mased amino acid (which encompasses mathematical calculations of repeatedly adjusting current values of the model parameters to determine trained values of the model parameters from the initial values of the model parameters using stochastic gradient descent and backpropagation which is a series of mathematical calculations see instant disclosure [0034], [0049], and [0065]), for each of the plurality of proteins, determining a confidence estimate for the target structure parameters for the protein (which encompasses calculating a difference between probability distributions which is a mathematical calculation see instant disclosure [0071]), training the structure prediction neural network to, for one or more other proteins, process representation of a multiple sequence alignment for the other protein to generate structure parameters that match ground truth parameters for the other protein (which encompasses mathematical calculations of repeatedly adjusting current values of the model parameters to determine trained values of the model parameters from the initial values of the model parameters using stochastic gradient descent and backpropagation which is a series of mathematical calculations see instant disclosure [0034], [0049], and [0065]), and determining a structure of a protein (which encompasses calculating numerical coordinates for a protein sequence using statistical features derived from a MSA see instant disclosure [0039]-[0040]). The MPEP states that “There is no particular word or set of words that indicates a claim recites a mathematical calculation. That is, a claim does not have to recite the word "calculating" in order to be considered a mathematical calculation. For example, a step of "determining" a variable or number using mathematical methods or "performing" a mathematical operation may also be considered mathematical calculations when the broadest reasonable interpretation of the claim in light of the specification encompasses a mathematical calculation” (see MPEP 2106.04(a)(2)(I)(C)).
Dependent claims 2-6, 12, 13, 15, 25 further limit the mental process/mathematical concept recited in the independent claim but do not change their nature as a mental process/mathematical concept.
(Step 2A Prong 2)
Claims found to recite a judicial exception under Step 2A, Prong 1 are then further analyzed to determine if the claims as a whole integrate the recited judicial exception into a practical application or not (Step 2A, Prong 2). Integration into a practical application is evaluated by identifying whether there are any additional elements recited in the claim and evaluating those additional elements to determine whether they integrate the exception into a practical application.
The additional element in claims 1, 9, 26, 34, and 36 using a neural network to perform the judicial exceptions does not integrate the judicial exceptions into a practical application because this additional element constitutes as mere instructions to apply the abstract idea to a neural network/computer environment (see MPEP 2106.05(f) and Example 47). This additional element constitutes as mere instructions to apply because the claims do not recite details of how the neural network functions (i.e., how the neural network processes a representation of the full multiple sequence alignments to generate output structure parameters). This additional element further constitutes as generally linking the judicial exception of processing a representation of the full multiple sequence alignments to generate output structure parameters to the technological environment of neural networks (see MPEP 2106.05(h)).
The additional element in claim 26 of obtaining an amino acid sequence of a protein and the additional element in claim 27 of extracting a protein from a human or animal body and obtaining the amino acid sequence from the extracted protein does not integrate the judicial exception into a practical application because these steps constitutes as insignificant extra solution activity of data gathering (see MPEP 2106.05(g)). These additional elements only interact with the judicial exceptions in a manner by providing data to the judicial exceptions to analyze.
The additional elements in claim 34 of one or more computers and one or more storage devices coupled to the computers which store instructions that, when executed by the one or more computers to perform the judicial exceptions and the additional element in claim 36 of one or more non-transitory computer storage media storing instructions that when executed by one or more computers cause the one or more computer to perform the judicial exceptions does not integrate the judicial exceptions into a practical application because this is applying the judicial exceptions to a generic computer/computer environment without an improvement to computer technology (see MPEP 2106.04(d)(1)). The generic computer/computer environment only interacts with the recited judicial exceptions in a manner that the computer is invoked as a tool to perform the judicial exceptions which does not constitute as an improvement to computer capabilities (see MPEP 2106.05(a)(I)).
Thus, the additional elements do not integrate the judicial exceptions into a practical application and claims 1-15, 25-27, 34, and 36 are directed to the abstract idea.
(Step 2B)
Claims found to be directed to a judicial exception are then further evaluated to determine if the claims recite an inventive concept that provides significantly more than the judicial exception itself (Step 2B). The claims do not include additional elements that are sufficient to amount to significantly more than the judicial exception because:
The additional element in claims 1, 9, 26, 34, and 36 using a neural network to perform the judicial exceptions (which constitutes as mere instructions to apply the judicial exception to a generic computing environment), the additional elements in claim 34 of one or more computers and one or more storage devices coupled to the computers which store instructions that, when executed by the one or more computers to perform the judicial exceptions, and the additional element in claim 36 of one or more non-transitory computer storage media storing instructions that when executed by one or more computers cause the one or more computer to perform the judicial exceptions is conventional as shown by MPEP 2106.05(b) and MPEP 2106.05(d)(II). Further, the additional element which limits the structure of the neural network configured to process a network input that comprises both: (i) a representation of a multiple sequence alignment for a protein, and (ii) a representation of an amino acid sequence of the protein (in claim 9) is conventional as shown by Yang et al. (Proc. Natl. Acad. Sci. U.S.A. 117 (3) 1496-1503, (2020)) on page 1502 left col., Senior et al. (WO 2020058176 A1) paragraph [0030], and Senior et al. (Nature, January 2020, 577 (7792): 706-710 (plus 17 supplement pages); cited in IDS received 24 October 2024) on page 6 of the reference (left col.).
The additional element in claim 26 of obtaining an amino acid sequence of a protein and the additional element in claim 27 of extracting a protein from a human or animal body and obtaining the amino acid sequence from the extracted protein is conventional as shown by the [0018] of the instant disclosure of obtaining a structure of a version of the protein obtained from a human or animal body, e.g., by conventional (physical) methods, such as X-ray crystallography, NMR spectroscopy or electron microscopy, [0114] of Senior et al. (WO 2020058176 A1), and Stollar et al. (Essays Biochem 8 October 2020; 64 (4): 649–680) which reviews physical methods for analyzing extracted proteins.
Thus, the additional elements are not sufficient to amount to significantly more than the judicial exception because they are conventional.
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.
Claims 1, 2, 8-10, 14, 15, 25-27, 34, and 36 are rejected under 35 U.S.C. 103 as being unpatentable over Senior et al. (WO 2020058176 A1) in view of Xie et al. (2020 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), Seattle, WA, USA, 2020, pp. 10684-10695).
Claim 1 is directed to a method performed by one or more data processing apparatus for training a structure prediction neural network that is configured to generate structure parameters that characterize a structure of a protein by processing a network input that comprises a representation of a multiple sequence alignment for the protein, the method comprising: obtaining, for each of a plurality of proteins, a full multiple sequence alignment for the protein
Senior et al. shows training a neural network which predicts distance distributions between amino acid pairs and torsion angle parameters of a protein which are interpreted as being structure parameters that characterize a structure of a protein through distances and torsion angles using features derived from a multiple sequence alignment (Senior et al. [0025], [0030] and [0031]). Senior et al. shows using an alignment algorithm to produce a multiple sequence alignment (MSA) for an amino acid sequence to generate alignment features (Senior et al. [0203]).
generating, for each of the plurality of proteins, target structure parameters characterizing a structure of the protein from the full multiple sequence alignment for the protein, comprising: processing a representation of the full multiple sequence alignment for the protein using the structure prediction neural network to generate output structure parameters characterizing a structure of the protein and determining the target structure parameters for the protein based on the output structure parameters for the protein
Senior shows a process of training a neural network to generate distance maps by using a teacher neural network to generate a target distance map, which is interpreted as being target structure parameters, by processing training inputs (Senior et al. [0241], claim 10, and claim 26). Senior et al. shows the neural network input is includes alignment features determined from the multiple sequence alignment for the amino acid sequence (Senior [0232] - [0235]).
determining, for each of the plurality of proteins, a reduced multiple sequence alignment for the protein, comprising removing or masking data from the full multiple sequence alignment for the protein,
Senior et al. shows augmenting the training data to generate new training examples where the alignment features are generated by performing random subsampling of on a MSA for the amino acid sequence (Senior et al. [0242]). This random subsampling is interpreted as removing data from the full multiple sequence alignment for the protein by randomly selecting a subsample of amino acid sequences while the other non-selected amino acid sequences are not used (e.g., removed) to generate the alignment features.
training the structure prediction neural network to, for one or more of the plurality of proteins, process a representation of the reduced multiple sequence alignment for the protein to generate structure parameters that match the target structure parameters for the protein.
Senior et al. shows training the neural network to generate outputs matching the outputs of a teacher neural network as a form of distillation learning which allows the neural network to be trained more effectively (Senior et al. [0241]).
Senior et al. does show that the generation of the target structure parameters are based on the full multiple sequence alignment (non-augmented/unreduced MSA) and training the structure prediction neural network to, for one or more of the plurality of proteins, process a representation of the reduced multiple sequence alignment for the protein (the augmented MSA).
Like Senior et al., Xie et al. shows a teacher-student learning framework for neural networks. Xie et al. shows training the student model with noisy data (i.e., augmented data) to match the output of the teacher model that infers output of unpaired training examples (Xie et al. page 10685 left col. - right col.). Xie et al. shows that the teacher network intakes clean data (e.g., data which has not been subjected to data augmentation) to produce a high-quality output while the student is required to reproduce the output of the teacher using augmented data as input (Xie et al. page 10685 right col.). Xie et al. shows that the data augmentation forces the student to ensure prediction consistency across augmented versions of the input data (Xie et al. page 10685 right col.).
Claim 34 is directed to a system comprising: one or more computers; and one or more storage devices communicatively coupled to the one or more computers, wherein the one or more storage devices store instructions that, when executed by the one or more computers, cause the one or more computers to perform operations for training a structure prediction neural network that is configured to generate structure parameters that characterize a structure of a protein by processing a network input that comprises a representation of a multiple sequence alignment for the protein by implementing the steps set out in method claim 1. Claim 36 is directed to one or more non-transitory computer storage media storing instructions that when executed by one or more computers cause the one or more computers to perform operations for training a structure prediction neural network that is configured to generate structure parameters that characterize a structure of a protein by processing a network input that comprises a representation of a multiple sequence alignment for the protein by implementing the steps set out in method claim 1.
Senior et al. shows the processes are implemented as computer programs on one or more computers (Senior et al. [0230]). Senior et al. shows the computer implemented processes can be implemented as computer programs encoded on a tangible non-transitory storage medium (Senior et al. [0384]).
Claim 2 is directed to wherein for each of the plurality of proteins, removing data from the full multiple sequence alignment for the protein comprises removing one or more amino acid sequences from the multiple sequence alignment for the protein.
Senior et al. shows augmenting the training data to generate new training examples where the alignment features are generated by performing random subsampling of the full MSA for the amino acid sequence (Senior et al. [0242]). This random subsampling is interpreted as removing data from the full multiple sequence alignment for the protein by randomly selecting a subsample of amino acid sequences while the other non-selected amino acid sequences are not used (e.g., removed) to generate the alignment features.
Claim 8 is directed to wherein determining the target structure parameters for the protein based on the output structure parameters for the protein comprises adding random noise values to the output structure parameters for the protein.
Senior et al. shows that for training examples random noise is added to the target distance map (Senior et al. claim 11 and claim 26).
Claim 9 is directed to wherein the structure prediction neural network is configured to process a network input that comprises both: (i) a representation of a multiple sequence alignment for a protein, and (ii) a representation of an amino acid sequence of the protein.
Senior et al. shows the network input may include extracting components of (i) a representation of the sequence of amino acid residues and (ii) alignment features derived from a multiple sequence alignment (MSA) which include covariation features amongst the sequences in the MSA which can help to identify residues in contact (Senior et al. [0030]).
Claim 10 is directed to further comprising, for each of the plurality of proteins, determining a confidence estimate for the target structure parameters for the protein.
Senior et al. shows determining a distance likelihood score which conveys more precise information about how closely the predicted structure conforms with the actual structure of the amino acid sequence (Senior et al. [0252]).
Claim 14 is directed to training the structure prediction neural network to, for one or more other proteins, process a representation of a multiple sequence alignment for the other protein to generate structure parameters that match ground truth structure parameters for the other protein. Claim 15 is directed to wherein the ground truth structure parameters for the other proteins are determined by physical experiments.
Senior et al. shows a training process of training a distance prediction neural network using training data including multiple training examples where each training example includes a training network input and a target distance map corresponding to the training network input (Senior et al. [0241]). Senior et al further shows that the target distance map characterizes the actual distances between amino acid residues in the structure of the training protein and that actual structures of different proteins that are stored in a structure database may have been determined using physical experimental methods such as x-ray crystallography (Senior et al. [0183] and [0241]).
Claim 25 is directed to wherein the structure parameters comprise one or both of a plurality of torsion angles and a plurality of atom coordinates.
Senior et al. shows the distance prediction neural network predicts torsion angle parameters (Senior et al. [0030]).
Claim 26 is directed to obtaining an amino acid sequence of a protein and using the trained structure prediction neural network to determine a structure of the protein. Claim 27 is directed to extracting the protein from a human or animal body and obtaining the amino acid sequence from the extracted protein.
Senior et al. shows obtaining an amino acid sequence of a protein and using a distance neural network to predict structural parameters of distances of residue pairs and torsion angles of the amino acid sequence of the protein to determine a structure of the protein (Senior et al. [0114]). Senior et al. shows obtaining a structure of a version of the protein obtained from a human or animal body by conventional (physical) methods (Senior et al. [0114]).
An invention would have been obvious to one or ordinary skill in the art if some motivation in the prior art would have led that person to modify reference teachings to arrive at the claimed invention. It would have been obvious to one of ordinary skill in the art before the effective filling date of the invention to have modified the neural network training that utilizes a teacher network to output a target distance map output which is used in training the neural network to generate distance maps using MSA features of Senior et al. with the teacher-student learning framework in which the teacher model intakes clean data to generate a high-quality output which is used in training a student model by requiring the student model to reproduce the output of the teacher using augmented data as input of Xie et al. because this would allow for a training process in which the teacher model intakes MSA features derived from clean data (i.e., non-augmented data) to produce a high quality target distance map output while requiring the distance prediction neural network to match the target distance map output using augmented data (e.g., produced by subsampling sequences in the full MSA) as input to ensure prediction consistency across augmented versions of data (Xie et al. page 10685 right col.). One would have a reasonable expectation of success because Senior et al. shows processes for performing data augmentation on multiple sequence alignments and a process for training a neural network which utilizes a teacher network to generate a target output used in training the neural network while Xie et al. shows a particular learning framework which implements a teacher neural network to generate an output and requiring a student neural network to match the output generated by the teacher using augmented data as input.
Claims 5 and 6 are rejected under 35 U.S.C. 103 as being unpatentable over Senior et al. in view of Xie et al. as applied to claim 1 above, and further in view of Kandathil et al. (Proteins. 2019 Dec;87(12):1092-1099).
Claim 5 is directed to masking an identity of a respective amino acid at one or more positions in one or more amino acid sequences in the full multiple sequence alignment for the protein. masking an identity of a respective amino acid at one or more positions in one or more amino acid sequences in the full multiple sequence alignment for the protein. Claim 6 is directed to randomly sampling the positions to be masked in the amino acid sequences in the full multiple sequence alignment for the protein.
Like Senior et al. in view of Xie et al., Kandathil et al. shows performing data augmentation to generate synthetic samples in protein structure predictions. Kandathil et al. shows a loop sampling process by masking or deleting residues in loops to generate synthetic examples to be used as training examples (Kandathil et al. page 1093 right col.). Kandathil et al. shows residues in bends are considered for removal with a probability of 0.3 and simulating deletions in loops by probabilistically removing residues classified in loops (Kandathil et al. page 1093 right col.). It is interpreted the removal with a probability of 0.3 and probabilistically removing residues includes removing the data by masking. The probabilistic removal process is interpreted as performing a random sampling of a distribution and removing residues when the probability sampled meets a probability threshold of 0.3.
An invention would have been obvious to one or ordinary skill in the art if some motivation in the prior art would have led that person to modify reference teachings to arrive at the claimed invention. It would have been obvious to one of ordinary skill in the art before the effective filling date of the invention to have modified the data augmentation process of Senior et al. in view of Xie et al. to include the data augmentation process of loop sampling which masks residues in loops to generate synthetic examples to be used as training examples of Kandathil et al. because this would allow for an additional data augmentation process which produces a variety of types of synthetic examples to be used in the training process of the neural network (Kandathil et al. page 1093 right col.). One would have a reasonable expectation of success because Senior et al. in view of Xie et al. shows data augmentation processes on multiple sequence alignment data to produce alignment features while Kandathil et al. shows performing loop sampling as a form of data augmentation on input tensors which are multiple sequence alignments generated form alignment algorithms.
Conclusion
No claims are allowed.
Claims 3, 4, 7 and 11-13 are free of the prior art of record. Senior et al. (WO 2020058176 A1) is the closest art of record which shows training a neural network to predict structure parameters of a protein sequence utilizing multiple sequence alignment features and data augmentation processes through subsampling multiple sequence alignments. However, the prior art of record does not show the limitations of: performing the particular process of “removing amino acid sequences by sampling a reduction parameter value from a set of possible reduction parameter values… wherein the reduction parameter value specifies a number of amino acid sequences to be removed from the full multiple sequence alignment and removing the specified number of amino acid sequences” (in claims 3-4), performing the particular training process of the neural network to generate an auxiliary output that predicts the identity of each masked amino acid in the reduced multiple sequence alignment (in claim 7), performing the process of identifying one or more proteins which the confidence estimate for the target structure parameters for the protein does not satisfy a threshold and refraining from training the structure prediction neural network on the identified proteins (in claim 11), performing the process of determining gradients of an objective function that measures an error wherein the error is scaled by a function of the confidence estimate for the target structure parameters for the protein (in claim 12), or the particular architecture of the neural network in which the confidence estimate for the target structure parameters for the protein is generated as an auxiliary output of the structure prediction neural network (in claim 13). Thus, claims 3, 4, 7, and 11-13 are free from the prior art of record.
This Office action is a Non-Final action. A shortened statutory period for reply to this action is set to expire THREE MONTHS from the mailing date of this action.
Any inquiry concerning this communication or earlier communications from the examiner should be directed to JONATHAN EDWARD HAYES whose telephone number is (571)272-6165. The examiner can normally be reached M-F 9am-5pm.
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.
/JONATHAN EDWARD HAYES/Examiner, Art Unit 1685
Prosecution Insights