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 .
Status of the Claims
Claims 1-20 are pending and under consideration in this action.
Priority
The instant application claims domestic benefit to U.S. Provisional Application No. 63/344,081, filed 05/20/2022, as reflected in the filing receipt mailed 06/08/2023. The claim for domestic benefit for claims 1-20 is acknowledged. As such, the effective filing date of claims 1-20 is 05/20/2022.
Information Disclosure Statement
The information disclosure statement (IDS) submitted on 05/18/2023 is in compliance with the provisions of 37 CFR 1.97. Accordingly, the IDS has been considered by the examiner.
It is further noted that certain references lack appropriate volume number, issue number, and/or article number (NPL #1, 4-8, 10, 12-16, and 20-23). The Examiner has annotated those references herein. Applicant is kindly reminded to provide proper citations in compliance with 37 CFR 1.97 in all future submissions to the office.
Drawings
The drawings are objected to as failing to comply with 37 CFR 1.84(p)(4) because:
Reference character “306” has been used to designate both “creating a vaccine” (Fig. 3 and Specification Para. [0054]) and “the binding motif” (Specification Para. [0054]).
Corrected drawing sheets in compliance with 37 CFR 1.121(d) are required in reply to the Office action to avoid abandonment of the application. Any amended replacement drawing sheet should include all of the figures appearing on the immediate prior version of the sheet, even if only one figure is being amended. Each drawing sheet submitted after the filing date of an application must be labeled in the top margin as either “Replacement Sheet” or “New Sheet” pursuant to 37 CFR 1.121(d). If the changes are not accepted by the examiner, the applicant will be notified and informed of any required corrective action in the next Office action. The objection to the drawings will not be held in abeyance.
Claim Objections
Claims 2 and 12 are objected to because of the following informalities:
Claim 2 recites the limitation “wherein calculating the binding motif of the MHC includes comprising generating a plurality of binding motifs for a plurality of respective MHCs”, which should be corrected to include either “includes” or “comprising”, not both, for clarity.
Claim 2 also recites the limitations “wherein calculating the binding motif of the MHC includes comprising generating a plurality of binding motifs for a plurality of respective MHCs” and “wherein screening includes screening in accordance with the plurality of binding motifs”, which should be corrected to include an “and” between the two wherein limitations, for clarity.
Claim 12 recites the limitations “wherein the computer program further causes the processor to generate a plurality of additional binding motifs for a plurality of respective additional MHCs” and “wherein screening includes screening in accordance with the plurality of additional binding motifs”, which should be corrected to include an “and” between the two wherein limitations, for clarity.
Appropriate correction is required.
Claim Interpretation
Creating peptide-based vaccines as recited in claims 10 and 20 is interpreted as an in-silico activity because the peptide mutation policy neural network is used to generate the peptide-based vaccine. Specifically, following Fig. 2 and Speciation Para. [0033]-[0035], the method trains a peptide mutation policy (block 204) and then uses a scoring function and the trained peptide mutation policy to generate binding peptides (block 206). Subsequently, following Fig. 3 and Specification Para. [0053]-[0054], the generated peptides with specific binding motifs (from block 206) are used to generate a set of peptide vaccine candidates (block 302), and match scores for vaccine candidates (block 304). Based on the matching scores, a vaccine is created (block 306), e.g., by generating neoantigens that incorporate a selected peptide vaccine candidate. All steps are part of the computational pipeline, and therefore being interpreted as in-silico steps.
Claim Rejections - 35 USC § 112(a)
The following is a quotation of the first paragraph of 35 U.S.C. 112(a):
(a) IN GENERAL.—The specification shall contain a written description of the invention, and of the manner and process of making and using it, in such full, clear, concise, and exact terms as to enable any person skilled in the art to which it pertains, or with which it is most nearly connected, to make and use the same, and shall set forth the best mode contemplated by the inventor or joint inventor of carrying out the invention.
The following is a quotation of the first paragraph of pre-AIA 35 U.S.C. 112:
The specification shall contain a written description of the invention, and of the manner and process of making and using it, in such full, clear, concise, and exact terms as to enable any person skilled in the art to which it pertains, or with which it is most nearly connected, to make and use the same, and shall set forth the best mode contemplated by the inventor of carrying out his invention.
Claim 10 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 10 recites the limitation “administering the vaccine to prevent an illness”. The Specification (see Para. [0011] and [0054]) discloses that based on the matching scores, block 306 creates a vaccine by, e.g., generating neoantigens that incorporate a selected peptide vaccine candidate. Block 308 then administers the vaccine to prevent the illness. As this is the only recitation of administering the vaccine in the Specification, the claim language merely reiterates the Specification. The Specification does not provide any further details, steps, or examples of how the vaccine is administered for any illness. Accordingly, the disclosure is not commensurate with the written description scope of the claim.
Claim Rejections - 35 USC § 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 1-20 are rejected under 35 U.S.C. 112(b) or 35 U.S.C. 112 (pre-AIA ), second paragraph, as being indefinite for failing to particularly point out and distinctly claim the subject matter which the inventor or a joint inventor (or for applications subject to pre-AIA 35 U.S.C. 112, the applicant), regards as the invention.
Claims 1 and 11 recite the limitation “training a peptide mutation policy neural network using reinforcement learning that includes a peptide presentation score as a reward” and “generating a plurality of new peptides using the peptide mutation policy” in lines 2-4 and 5-7 of the claims, respectively. The metes and bounds of the claim are rendered indefinite due to the lack of clarity. The claims fail to particularly point out the steps required for the claimed invention functions. The claims specify the training of a peptide mutational policy neural network and generating a plurality of new peptides using the peptide mutation policy. However, it is unclear what data is input or what data is obtained as output such that generation would be achieved. Clarification through clearer claim language is respectfully requested. Claims 2-10 and 12-20 are also rejected due to their dependency on claims 1 and 11.
Claims 1 and 11 also recite the limitation “screening a plurality of library peptides in accordance with the binding motif” in lines 7 and 9 of the claims, respectively. The metes and bounds of the claims are rendered indefinite due to the lack of clarity. It is unclear what steps or parameters are required for the screening to be “in accordance with the binding motif”. Said differently, it is unclear what aspects or parameters of the binding motif are applied in the screening of the library peptides. It is also unclear what result is achieved when the screening of library peptides is performed, such as, for example, the identification of a candidate peptide from the library. Clarification through clearer claim language is respectfully requested. Claims 2-10 and 12-20 are also rejected due to their dependency on claims 1 and 11.
Claims 2 and 12 recite the limitation “wherein screening includes screening in accordance with the plurality of [additional] binding motifs”. The metes and bounds of the claims are rendered indefinite due to the lack of clarity. Analogous to claims 1 and 11 above, it is unclear what steps or parameters are required for the screening to be “in accordance with the plurality of binding motifs”. It is also unclear what result is achieved when the screening of the library peptides is performed. Clarification through clearer claim language is respectfully requested.
Claims 3 and 13 recite the phrase “wherein …
c
l
i
p
∙
is a clipping function …”. The metes and bounds of the claim are rendered indefinite due to the lack of clarity. It is unclear if the
c
l
i
p
∙
refers to the complete function,
c
l
i
p
(
r
t
θ
,
1
-
ϵ
,
1
+
ϵ
)
, as recited in the objective function equation. This rejection can be overcome by amendment of claims 3 and 13 to recite “wherein …
c
l
i
p
(
r
t
θ
,
1
-
ϵ
,
1
+
ϵ
)
,
is a clipping function …”.
Claim 10 recites the limitation “creating a vaccine based on the candidate vaccine peptide and administering the vaccine to prevent an illness”. The metes and bounds of the claim are rendered indefinite due to the lack of clarity. Claim 1, which is in the chain of dependency for claim 10, recites “a computer implemented method for peptide generation”. As described in the Claim Interpretation section above, the creation of the peptide-based vaccine is carried out in-silico. It is unclear what steps are required to administer an in silico vaccine to prevent an illness. Clarification through clearer claim language is respectfully requested.
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 1-20 are rejected under 35 U.S.C. 101 because the claimed invention is directed to an abstract idea without significantly more. The claims recite both (1) mathematical concepts (mathematical relationships, formulas or equations, or mathematical calculations) and (2) mental processes, i.e., concepts performed in the human mind (including observations, evaluations, judgements or opinions) (see MPEP § 2106.04(a)).
Framework with which to evaluate Subject Matter Eligibility as outlined in MPEP § 2106:
Step 1: Are the claims directed to a process, machine, manufacture or composition of matter;
Step 2A, Prong One: Do the claims recite a judicially recognized exception, i.e., a law of nature, a natural phenomenon, or an abstract idea;
Step 2A, Prong Two: If the claims recite a judicial exception under Prong One, then is the judicial exception integrated into a practical application (Prong Two); and
Step 2B: If the claims do not integrate the judicial exception, do the claims provide an inventive concept.
Framework as it pertains to the instant claims:
Step 1:
In the instant application, claims 1-10 are directed towards a method and claims 11-20 are directed towards a system, which falls into one of the categories of statutory subject matter (Step 1: YES).
Step 2A, Prong One:
In accordance with MPEP § 2106, claims found to recite statutory subject matter (Step 1: YES) are then analyzed to determine if the claims recite any concepts that equate to an abstract idea, law of nature or natural phenomenon (Step 2A, Prong One). The following instant claims recite limitations that equate to one or more categories of judicial exceptions:
Claims 1 and 11 recite a mathematical concept (i.e., training by maximizing an objective function for learning the mutation policy using the equation in Specification Para. [0043]) in “training a peptide mutation policy neural network using reinforcement learning that includes a peptide presentation score as a reward”; a mental process (i.e., evaluating the peptides to determine a binding motif) in “calculating a binding motif of a major histocompatibility complex (MHC) using the plurality of new peptides”; and a mental process (i.e., evaluating library peptides compared to the binding motif) in “screening a plurality of library peptides in accordance with the binding motif”.
Claims 2 and 12 recite a mental process (i.e., an evaluation of peptides to determine a binding motif and an evaluation of library peptides compared to the binding motif) in “wherein calculating the binding motif of the MHC includes comprising generating a plurality of binding motifs for a plurality of respective MHCs”.
Claims 3 and 13 recite a mathematical concept (i.e., maximizing a function) in “wherein training the peptide mutation policy neural network maximizes an objective function as:
max
θ
L
C
L
I
P
θ
=
E
^
t
[
min
(
r
t
(
θ
)
A
^
t
,
c
l
i
p
(
r
t
θ
,
1
-
ϵ
,
1
+
ϵ
)
A
^
t
)
]
, where
θ
represents parameters of the peptide mutation policy neural network,
E
^
t
is an expectation with respect to a time step t,
r
t
(
θ
)
is a probability ratio between an action under a current policy and an action under a previous policy,
A
^
t
is an average at time step t,
c
l
i
p
∙
is a clipping function, and
ϵ
is a size of a clipping interval”.
Claims 4 and 14 recite a mathematical concept (i.e., pretraining using the objective function to minimize cross-entropy loss in Specification Para. [0047]) in “wherein training the peptide mutation policy neural network includes pre-training using an expert policy”.
Claims 5 and 15 recite a mathematical concept (i.e., a calculation of pairwise distances) in “wherein screening the plurality of library peptides includes determining pairwise Euclidean distances between block substitution matrix (BLOSUM) representations of the plurality of library peptides and a BLOSUM representation of the binding motif”.
Claims 6 and 16 recite a mathematical concept (i.e., calculating log-likelihoods) in “wherein screening the plurality of library peptides includes determining log-likelihoods of the plurality of library peptides under a weighted position of the binding motif”.
Claims 8 and 18 recite a mental process (i.e., an evaluation of a changed input sequence and an evaluation of presentation score to determine a reward) in “wherein training the peptide mutation policy neural network includes changing an input peptide sequence as an action and determining a reward for the action based on the peptide presentation score of the changed input peptide sequence”.
Claims 9 and 19 recite a mental process (i.e., an evaluation of peptides compared to a candidate vaccine to determine binding) in “comparing the screened plurality of library peptides to a candidate vaccine peptide to determine how the candidate vaccine peptide binds to the MHC”.
Claims 10 and 20 recites a mental process (i.e., an evaluation of peptides to create a vaccine in-silico, see Claim Interpretation above) in “creating a vaccine based on the candidate vaccine peptide”.
These recitations are similar to the concepts of collecting information, and displaying certain results of the collection and analysis in Electric Power Group, LLC, v. Alstom (830 F.3d 1350, 119 USPQ2d 1739 (Fed. Cir. 2016)), comparing information regarding a sample or test to a control or target data in Univ. of Utah Research Found. v. Ambry Genetics Corp. (774 F.3d 755, 113 U.S.P.Q.2d 1241 (Fed. Cir. 2014)) and Association for Molecular Pathology v. USPTO (689 F.3d 1303, 103 U.S.P.Q.2d 1681 (Fed. Cir. 2012)), and organizing and manipulating information through mathematical correlations in Digitech Image Techs., LLC v Electronics for Imaging, Inc. (758 F.3d 1344, 111 U.S.P.Q.2d 1717 (Fed. Cir. 2014)) that the courts have identified as concepts that can be practically performed in the human mind or mathematical relationships.
The abstract ideas recited in the claims are evaluated under the broadest reasonable interpretation (BRI) of the claim limitations when read in light of and consistent with the specification, and are determined to be directed to mental processes that in the simplest embodiments are not too complex to practically perform in the human mind. Additionally, the recited limitations that are identified as judicial exceptions from the mathematical concepts grouping of abstract ideas are abstract ideas irrespective of whether or not the limitations are practical to perform in the human mind.
Specifically, claims 1 and 11 involve nothing more than training a peptide mutation policy, determining a binding motif, and screening library peptides with the binding motif. The step reciting training a peptide mutation policy is, under the BRI, performed using mathematical operations. The instant Specification (see Para. [0043]) discloses that learning the mutation policy is performed by maximizing the objective function using the specified equation. Additionally, since there are no specifics in the methodology, the steps reciting determining a binding motif and screening library peptides with the binding motif are something, that under the BRI, one could perform mentally. Therefore, the claimed steps are not further defined beyond something that reads on performing a calculation using a computer as a tool and merely looking at data and making a determination. As such, said steps are directed to judicial exceptions. The instant claims must therefore be examined further to determine whether they integrate the abstract idea into a practical application (Step 2A, Prong One: YES).
Step 2A, Prong Two:
In determining whether a claim is directed to a judicial exception, further examination is performed that analyzes if the claim recites additional elements that when examined as a whole integrates the judicial exception(s) into a practical application (MPEP § 2106.04(d)). A claim that integrates a judicial exception into a practical application will apply, rely on, or use the judicial exception in a manner that imposes a meaningful limit on the judicial exception. The claimed additional elements are analyzed to determine if the abstract idea is integrated into a practical application (MPEP § 2106.04(d)(I)). If the claim contains no additional elements beyond the abstract idea, the claim fails to integrate the abstract idea into a practical application (MPEP § 2106.04(d)(III)). The following independent claims recite limitations that equate to additional elements:
Claim 1 recites “a computer implemented method” and “generating a plurality of new peptides using the peptide mutation policy”.
Claim 11 recites “a hardware processor”; “a memory that stores a computer program which, when executed by the hardware processor, causes the hardware processor to [perform steps]”; and “generating a plurality of new peptides using the peptide mutation policy”.
Regarding the above cited limitations in claims 1 and 11 of (i) a computer implemented method (claim 1); (ii) a hardware processor (claim 11); and (iii) a memory that stores a computer program which, when executed by the hardware processor, causes the hardware processor to [perform steps] (claim 11). These limitations require only a generic computer component, which does not improve computer technology. Therefore, these limitations equate to mere instructions to implement an abstract idea on a generic computer, which the courts have established does not render an abstract idea eligible in Alice Corp. 573 U.S. at 223, 110 USPQ2d at 1983.
Regarding the above cited limitations in claims 1 and 11 of (iv) generating a plurality of new peptides using the peptide mutation policy. These limitations equate to insignificant, extra-solution activity of mere data gathering because these limitations gather data before or after the recited judicial exceptions of calculating a binding motif of a major histocompatibility complex and screening library peptides with the binding motif (see MPEP § 2106.04(d)).
Additionally, none of the recited dependent claims recite additional elements which would integrate the judicial exception into a practical application. Specifically, claims 7 and 17 further limit the generation of new peptides using the peptide mutation policy. As such, claims 1-20 are directed to an abstract idea (Step 2A, Prong Two: NO).
Step 2B:
Claims found to be directed to a judicial exception are then further evaluated to determine if the claims recite an inventive concept that provides significantly more than the judicial exception itself (Step 2B). The claims do not include additional elements that are sufficient to amount to significantly more than the judicial exception. The instant independent claims recite the same additional elements described in Step 2A, Prong Two above.
Regarding the above cited limitations in claims 1 and 11 of (i) a computer implemented method (claim 1); (ii) a hardware processor (claim 11); and (iii) a memory that stores a computer program which, when executed by the hardware processor, causes the hardware processor to [perform steps] (claim 11). These limitations equate to instructions to implement an abstract idea on a generic computing environment, which the courts have established does not provide an inventive concept (see MPEP § 2106.05(d) and MPEP § 2106.05(f)).
Regarding the above cited limitations in claims 1 and 11 of (iv) generating a plurality of new peptides using the peptide mutation policy. These limitations when viewed individually and in combination, are WURC limitations as taught by Skwark et al. (Designing a Prospective COVID-19 Therapeutic with Reinforcement Learning. arXiv:2012.01736 [q-bio.BM] (12 pages) (2020)). Skwark et al. discloses a novel protein design framework as a reinforcement learning (RL) problem. They generate new designs efficiently through the combination of a fast, biologically-grounded reward function and sequential action-space formulation (Abstract). Skwark et al. further teaches that they used their method to design a novel variant of the human angiotensin-converting enzyme 2 (ACE2) that binds more tightly to the SARS-COV-2 spike protein and diverts it from human cells (Abstract). They used the RL method to produce candidate designs above the threshold of the native ACE2 in a much more efficient manner (limitation (iv)) (Pg. 3, Para. 6).
These additional elements do not comprise an inventive concept when considered individually or as an ordered combination that transforms the claimed judicial exception into a patent-eligible application of the judicial exception. Therefore, the instant claims do not amount to significantly more than the judicial exception itself (Step 2B: NO). As such, claims 1-20 are not patent eligible.
Claim Rejections - 35 USC § 103
In the event the determination of the status of the application as subject to AIA 35 U.S.C. 102 and 103 (or as subject to pre-AIA 35 U.S.C. 102 and 103) is incorrect, any correction of the statutory basis (i.e., changing from AIA to pre-AIA ) for the rejection will not be considered a new ground of rejection if the prior art relied upon, and the rationale supporting the rejection, would be the same under either status.
The following is a quotation of 35 U.S.C. 103 which forms the basis for all obviousness rejections set forth in this Office action:
A patent for a claimed invention may not be obtained, notwithstanding that the claimed invention is not identically disclosed as set forth in section 102, if the differences between the claimed invention and the prior art are such that the claimed invention as a whole would have been obvious before the effective filing date of the claimed invention to a person having ordinary skill in the art to which the claimed invention pertains. Patentability shall not be negated by the manner in which the invention was made.
The factual inquiries for establishing a background for determining obviousness under 35 U.S.C. 103 are summarized as follows:
1. Determining the scope and contents of the prior art.
2. Ascertaining the differences between the prior art and the claims at issue.
3. Resolving the level of ordinary skill in the pertinent art.
4. Considering objective evidence present in the application indicating obviousness or nonobviousness.
This application currently names joint inventors. In considering patentability of the claims the examiner presumes that the subject matter of the various claims was commonly owned as of the effective filing date of the claimed invention(s) absent any evidence to the contrary. Applicant is advised of the obligation under 37 CFR 1.56 to point out the inventor and effective filing dates of each claim that was not commonly owned as of the effective filing date of the later invention in order for the examiner to consider the applicability of 35 U.S.C. 102(b)(2)(C) for any potential 35 U.S.C. 102(a)(2) prior art against the later invention.
1. Claims 1-2, 7-8, 11-12, and 17-18 are rejected under 35 U.S.C. 103 as being unpatentable over Skwark et al. (Designing a Prospective COVID-19 Therapeutic with Reinforcement Learning. arXiv:2012.01736 [q-bio.BM] (12 pages) (2020); published 12/03/2020) in view of Chen et al. (Ranking-Based Convolutional Neural Network Models for Peptide-MHC Class I Binding Prediction. Front. Mol. Biosci. 8: 634836 (17 pages) (2021); published 05/16/2021).
Regarding claim 1, Skwark et al. teaches a novel protein design framework as a reinforcement learning (RL) problem. They generate new designs efficiently through the combination of a fast, biologically-grounded reward function and sequential action-space formulation (Abstract). Skwark et al. further teaches that protein design is expressed as a Markovian Decision Process in Fig. 1. The figure shows the iterative problem formulation, where the first state is initialized with a sequence to be mutated by the RL agent. At each timestep, a position/amino acid is selected for mutation, and the process repeats until the agent stops, thereby outputting a final sequence (Pg. 2, Fig. 1). They formulate the design of high affinity binder proteins as a Markov Decision Process (MDP) (
S
,
A
,
T
,
R
,
γ
), where
S
is the state space,
A
is the action space,
T
:
S
×
A
→
S
is the transition function,
R
is the reward signal, and
γ
is a discount factor. One episode corresponds to the design of a protein. An environment state represents a protein, i.e. a sequence of amino acids
s
=
(
s
1
,
s
2
,
…
,
s
n
)
, where
s
i
refers to the residue at position
i
and
n
is the length of the complete sequence. A possible formulation is to initialize each episode with a blank sequence and then generate each amino acid progressively. In this case, the agent chooses
n
times between
m
actions during each episode. At the end of an episode, the final protein obtained is scored. The resulting reward signal is sparse, i.e non-zero only at the end of an episode. In the iterative formulation, the agent begins each episode with an initial sequence. The agent then mutates the sequence by selecting positions to change, as well as the new residues to change them to, in such a way that there is always a valid sequence to be scored. As a result, the mutation process can be stopped at any moment, potentially by the agent itself (Pg. 2, Para. 3-4). Skwark et al. further teaches the training of the policy (Pg. 3, Para. 6 and Pg. 8, Table 1), as well as the neural network architecture (Pg. 7, Para. 2 and Pg. 7, Fig. 3) (i.e., training a peptide mutation policy neural network using reinforcement learning that includes a peptide presentation score as a reward). Skwark et al. further teaches that they used their method to design a novel variant of the human angiotensin-converting enzyme 2 (ACE2) that binds more tightly to the SARS-COV-2 spike protein and diverts it from human cells (Abstract). They used the RL method to produce candidate designs above the threshold of the native ACE2 in a much more efficient manner (Pg. 3, Para. 6). They compared the best scoring designs from their method to previously report designs by constructing a pool of structural models. They showed that their designs both bind ACE2 and retain sufficient similarity to human ACE2 receptor, to evade recognition by immune system, and to make them plausible as drop-in therapeutics (i.e., generating a plurality of new peptides using the peptide mutation policy) (Pg. 4, Para. 1-2).
Regarding claim 7, Skwark et al. teaches that the agent begins each episode with an initial sequence. The agent then mutates the sequence by selecting positions to change, as well as the new residues to change them to, in such a way that there is always a valid sequence to be scored. As a result, the mutation process can be stopped at any moment, potentially by the agent itself. The choice of initial sequence can be arbitrary, though we choose either the native human ACE2 protein sequence, or randomly perturbed versions thereof (i.e., wherein generating the plurality of new peptides includes sampling a random starting peptide and applying a change to the random starting peptide according to the peptide mutation policy) (Pg. 2, Para. 4).
Regarding claim 8, Skwark et al. teaches the training of the policy (Pg. 3, Para. 6 and Pg. 8, Table 1). Skwark et al. further teaches that each episode is initialized with a blank sequence and then generate each amino acid progressively. In this case, the agent chooses
n
times between
m
actions during each episode. At the end of an episode, the final protein obtained is scored. The resulting reward signal is sparse, i.e non-zero only at the end of an episode (i.e., wherein training the peptide mutation policy neural network includes changing an input peptide sequence as an action and determine a reward for the action based on the peptide presentation score of the changed input peptide sequence) (Pg. 2, Para. 3).
Regarding claim 11, Skwark et al. teaches that the use of Policy Gradients reduces the compute budget needed to reach consistent, high-quality designs by at least an order of magnitude compared to standard methods (i.e., carried out on a computer containing a hardware processor and a memory that stores a computer program, which when executed by the hardware processor, causes the hardware processor to [perform steps]) (Abstract). Skwark et al. further teaches the limitations of train a peptide mutation policy neural network using reinforcement learning that includes a peptide presentation score as a reward and generate a plurality of new peptides using the peptide mutation policy as described for claim 1 above.
Regarding claim 17, Skwark et al. teaches the limitation of wherein the computer program further causes the processor to sample a random starting peptide and applying a change to the random starting peptide according to the peptide mutation policy as described for claim 7 above.
Regarding claim 18, Skwark et al. teaches the limitation of wherein the computer program further causes the processor to change an input peptide sequence as an action and to determine a reward for the action based on the peptide presentation score of the changed input peptide sequence as described for claim 8 above.
Skwark et al. does not teach calculating a binding motif of a major histocompatibility complex (MHC) using the plurality of new peptides (claims 1 and 11); screening a plurality of library peptides in accordance with the binding motif (claims 1 and 11); and wherein calculating the binding motif of the MHC includes comprising generating a plurality of binding motifs for a plurality of respective MHCs, wherein screening includes screening in accordance with the plurality of binding motifs (claims 2 and 12).
Regarding claims 1 and 11, Chen et al. teaches a method for identifying peptides that can bind to MHC class-I molecules, as this plays a vital role in the design of peptide vaccines. They develop two allele-specific convolutional neural network based methods to tackle the binding prediction problem by optimizing the rankings of peptide-MHC bindings (Abstract). Chen et al. further teaches that they used a combined dataset that consists of 202,510 entries across 128 alleles and 53,253 peptides (Pg. 3, Col. 2, Para. 3). Chen et al. further teaches allele-specific binding motifs of the peptides with high affinity for alleles HLA-A*02:01 and HLA-A*24:02 in Figure 3 (Pg. 11, Fig. 3 and Pg. 12, Col. 2, Para. 2). For HLA-A*02:01, the amino acids located at the second position and the last position contribute most to the binding events. This is consistent with the conserved motif calculated by stabilized matrix method (SMM) matrix. For HLA-A*24:02, the second and last positions have higher weights (i.e., calculating a binding motif of a major histocompatibility complex (MHC) using the plurality of new peptides) (Pg. 12, Col. 1, Para. 2 – Col. 2, Para. 1). Chen et al. further teaches that they evaluated the performance of their methods with the Mass Spectrometry benchmark dataset curated by O’Donnell et al. (2018). This MS benchmark dataset contains 23,653 sequences of MHC-displayed ligands eluted from B cell lines expressing 15 MHC class I alleles. For each eluted ligand, 100 decoys will be sampled from the protein-coding transcripts that contained this eluted ligand. Specifically, they sampled an equal number of decoys of each length 8–15. After removing all the entries present in the Immune Epitope Database (IEDB) dataset, the yielded Mass Spectrometry benchmark dataset contains 23,653 positive peptides and 2,377,037 randomly sampled negative peptides. Their ensemble methods achieve either the best or the second best performance on 12 out of 15 alleles among all the methods. When their ensemble achieves the second best performance on an allele, it is very comparable to the best performance—on average, the difference is 0.0112. For HLA-B alleles, our ensemble methods are also the best or the second best (i.e., screening a plurality of library peptides in accordance with the binding motif) (Pg. 15, Col. 1, Para. 4 and Pg. 15, Col. 2, Para. 2).
Regarding claims 2 and 12, Chen et al. teaches the binding motifs for HLA-A*02:01 and HLA-A*24:02 in Figure 3 and HLA-B*27:05 and HLA-B*58:01 in Fig. 4 (i.e., wherein calculating the binding motif of the MHC includes comprising generating a plurality of binding motifs for a plurality of respective MHCs) (Pg. 11, Fig 3 and Pg. 12, Fig. 4). Chen et al. further teaches that the Mass Spectrometry benchmark data set includes 15 MHC class I alleles (i.e., wherein screening includes screening in accordance with the plurality of binding motifs) (Pg. 15, Col. 1, Para. 4).
Therefore, regarding claims 1-2, 7-8, 11-12, and 17-18, it would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to modify the method of using reinforcement learning to generate novel protein designs of Skwark et al. with the analysis of binding motifs for MHC proteins of Chen et al. because the method of Skwark et al. can accelerate the design of peptide-based therapeutics for numerous diseases (Skwark et al., Abstract). Additionally, the method of Chen et al. can be used to identify peptides that can bind to MHC-class I molecules, and play a vital role in the design of peptide vaccines (Chen et al., Abstract) One of ordinary skill in the art would be able to combine the teachings of Skwark et al. with Chen et al. with reasonable expectation of success due to the same nature of the problem to be solved, since both incorporate a method for using deep learning to design peptide therapeutics. Therefore, regarding claims 1-2, 7-8, 11-12, and 17-18, the instant invention is prima facie obvious (MPEP § 2142).
2. Claims 3 and 13 are rejected under 35 U.S.C. 103 as being unpatentable over Skwark et al. in view of Chen et al., as applied to claims 1-2, 7-8, 11-12, and 17-18 above, and further in view of Schulman et al. (Proximal Policy Optimization Algorithms. arXiv:1707.06347v2 (12 pages) (2017); provided in the IDS dated 05/18/2023; published 08/28/2017).
Skwark et al. in view of Chen et al., as applied to claims 1-2, 7-8, 11-12, and 17-18 above, does not teach wherein training the peptide mutation policy neural network maximizes an objective function as:
max
θ
L
C
L
I
P
θ
=
E
^
t
[
min
(
r
t
(
θ
)
A
^
t
,
c
l
i
p
(
r
t
θ
,
1
-
ϵ
,
1
+
ϵ
)
A
^
t
)
]
, where
θ
represents parameters of the peptide mutation policy neural network,
E
^
t
is an expectation with respect to a time step t,
r
t
(
θ
)
is a probability ratio between an action under a current policy and an action under a previous policy,
A
^
t
is an average at time step t,
c
l
i
p
∙
is a clipping function, and
ϵ
is a size of a clipping interval.
Regarding claims 3 and 13, Schulman et al. teaches a new family of policy gradient methods for reinforcement learning, called proximal policy optimization (PPO), with a novel objective function that enables multiple epochs of minibatch updates (Abstract). Schulman et al. further teaches that trust region policy optimization (TRPO) maximizes a "surrogate" objective in the following formula:
L
C
L
I
P
θ
=
E
^
t
[
min
(
r
t
(
θ
)
A
^
t
,
c
l
i
p
(
r
t
θ
,
1
-
ϵ
,
1
+
ϵ
)
A
^
t
)
]
, where
r
t
(
θ
)
is a probablility ratio,
r
t
θ
=
π
θ
(
a
t
|
s
t
)
π
θ
,
o
l
d
(
a
t
|
s
t
)
, so
r
t
θ
o
l
d
=
1
;
A
^
t
is an estimator of the advantage function at timestep
t
; and the expectation
E
^
t
[
…
]
indicates the empirical average over a finite batch of samples. The second term, c
c
l
i
p
(
r
t
θ
,
1
-
ϵ
,
1
+
ϵ
)
A
^
t
, modifies the surrogate objective by clipping the probability ratio, which removes the incentive for moving
r
t
outside of the interval
[
1
-
ϵ
,
1
+
ϵ
]
. Finally, they take the minimum of the clipped and unclipped objective, so the final objective is a lower bound (a pessimistic bound) on the unclipped objective (i.e. wherein training the peptide mutation policy neural network maximizes an objective function as:
max
θ
L
C
L
I
P
θ
=
E
^
t
[
min
(
r
t
(
θ
)
A
^
t
,
c
l
i
p
(
r
t
θ
,
1
-
ϵ
,
1
+
ϵ
)
A
^
t
)
]
, where
θ
represents parameters of the peptide mutation policy neural network,
E
^
t
is an expectation with respect to a time step t,
r
t
(
θ
)
is a probability ratio between an action under a current policy and an action under a previous policy,
A
^
t
is an average at time step t,
c
l
i
p
∙
is a clipping function, and
ϵ
is a size of a clipping interval) (Pg. 2, Para 1-2 and Pg. 3, Para. 1-2).
Therefore, regarding claims 3 and 13, it would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to modify the method of using reinforcement learning to generate novel protein designs of Swartz et al. in view of Chen et al. with the objective function of Schulman et al. because the proximal policy optimization (PPO) outperforms either online policy gradient methods, and overall strikes a favorable balance between sample complexity, simplicity, and wall-time (Schulman et al., Abstract). Additionally, the method of Skwark et al. extends PPO to the sequential setting for the sequential action space configurations, thereby building on the PPO method of Schulman et al. (Skwark et al., Pg. 3, Para. 4). One of ordinary skill in the art would be able to combine the teachings of Swartz et al. in view of Chen et al. with Schulman et al. with reasonable expectation of success due to the same nature of the problem to be solved, since both implement the policy gradient method using PPO. Therefore, regarding claims 3 and 13, the instant invention is prima facie obvious (MPEP § 2142).
3. Claims 4 and 14 are rejected under 35 U.S.C. 103 as being unpatentable over Skwark et al. in view of Chen et al., as applied to claims 1-2, 7-8, 11-12, and 17-18 above, and further in view of Eysenbach et al. (Diversity is All you Need: Learning Skills without a Reward Function. arXiv:1802.06070v6 (22 pages) (2018); published 10/09/2018).
Skwark et al. in view of Chen et al., as applied to claims 1-2, 7-8, 11-12, and 17-18 above, does not teach wherein training the peptide mutation policy neural network includes pre-training using an expert policy.
Regarding claims 4 and 14, Eysenbach et al. teaches a method showing how pretrained skills can provide a good parameter initialization for downstream tasks, and can be composed hierarchically to solve complex, sparse reward tasks in reinforcement learning (RL) (Abstract). Eysenbach et al. further teaches that aside from maximizing reward with finetuning and hierarchical RL, they can also use learned skills to follow expert demonstrations. They consider the setting where they are given an expert trajectory consisting of states, without actions. Their goal is to obtain a feedback controller that will reach the same states. Given the expert trajectory, they use their learned discriminator to estimate which skill was most likely to have generated the trajectory (i.e., wherein training the peptide mutation policy neural network includes pre-training using an expert policy) (Pg. 8, Para. 3).
Therefore, regarding claims 4 and 14, it would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to modify the method of using reinforcement learning to generate novel protein designs of Swartz et al. in view of Chen et al. with the expert policy training of Eysenbach et al. because the results of Eysenbach et al. suggest that unsupervised discovery of skills can serve as an effective pretraining mechanism for overcoming challenges of exploration and data efficiency in reinforcement learning (Eysenbach et al., Abstract). One of ordinary skill in the art would be able to combine the teachings of Skwark et al. in view of Chen et al. with Eysenbach et al. with reasonable expectation of success due to the same nature of the problem to be solved, since incorporate a method for training in reinforcement learning. Therefore, regarding claims 4 and 14, the instant invention is prima facie obvious (MPEP § 2142).
4. Claims 5 and 15 are rejected under 35 U.S.C. 103 as being unpatentable over Skwark et al. in view of Chen et al., as applied to claims 1-2, 7-8, 11-12, and 17-18 above, and further in view of Song et al. (Parameterized BLOSUM Matrices for Protein Alignment. IEEE/ACM Trans Comput Biol Bioinform. 12(3): 686-94 (2015); published 10/31/2014).
Skwark et al. in view of Chen et al., as applied to claims 1-2, 7-8, 11-12, and 17-18 above, does not teach wherein screening the plurality of library peptides includes determining pairwise Euclidean distances between block substitution matrix (BLOSUM) representations of the plurality of library peptides and a BLOSUM representation of the binding motif.
Regarding claims 5 and 15, Song et al. teaches a method a for parameterizing BLOSUM matrices used in protein alignment (Title, Abstract). Song et al. further teaches that to apply the clustering, they need to calculate the distance between two aligned sequences. First, the distance between two amino acid is defined. A BLOSUM matrix
B
is normalized, such that all the elements of
B
'
are within [0,1]. Then the distance between two amino acid
x
,
y
is calculated as
d
x
,
y
=
∑
a
∈
A
B
'
x
,
a
-
B
'
y
,
a
2
/
|
A
|
, where
A
is the set of amino acids. Consequently, the distance between two aligned sequences
S
1
and
S
2
are calculated as
D
S
1
,
S
2
=
∑
i
d
(
S
1
i
,
S
2
i
)
(i.e., wherein screening the plurality of library peptides includes determining pairwise Euclidean distances between block substitution matrix (BLOSUM) representations of the plurality of library peptides and a BLOSUM representation of the binding motif) (Pg. 692, Col. 2, Para. 2).
Therefore, regarding claims 5 and 15, it would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to modify the method of using reinforcement learning to generate novel protein designs of Swartz et al. in view of Chen et al. with the calculation of BLOSUM matrices of Song et al. because Chen et al. discloses the use of BLOSUM matrices to represent peptides (Chen et al., Pg. 4, Col. 2, Para. 3). Song et al. builds upon the use of BLOSUM matrices, and enables alignment and accurate clustering of MHC II protein sequences (Song et al., Pg. 693, Col. 1, Para. 3). One of ordinary skill in the art would be able to combine the teachings of Skwark et al. in view of Chen et al. with Song et al. with reasonable expectation of success due to the same nature of the problem to be solved, since both incorporate a method for using BLOSUM matrices to represent proteins. Therefore, regarding claims 5 and 15, the instant invention is prima facie obvious (MPEP § 2142).
5. Claims 6 and 16 are rejected under 35 U.S.C. 103 as being unpatentable over Skwark et al. in view of Chen et al., as applied to claims 1-2, 7-8, 11-12, and 17-18 above, and further in view of Bassani-Sternberg et al. (Unsupervised HLA Peptidome Deconvolution Improves Ligand Prediction Accuracy and Predicts Cooperative Effects in Peptide–HLA Interactions. The Journal of Immunology. 197(6): 2492-2499 (2016); published 09/15/2016).
Skwark et al. in view of Chen et al., as applied to claims 1-2, 7-8, 11-12, and 17-18 above, does not teach wherein screening the plurality of library peptides includes determining log-likelihoods of the plurality of library peptides under a weighted position of the binding motif.
Regarding claims 6 and 16, Bassani-Sternberg et al. teaches a novel unsupervised approach to assign mass spectrometry–based HLA peptidomics data to their cognate HLA molecules. They show that incorporation of deconvoluted HLA peptidomics data in ligand prediction algorithms can improve their accuracy for HLA alleles with few ligands in existing databases (Abstract). Bassani-Sternberg et al. further teaches that Position weight matrices (PWMs) mathematically describing each allele’s binding motif were built by computing the frequency of each amino acid at each position, adding a random count of 1 at each position and each residue to account for low sampling. Sequence logos describing the motif of each allele (see Fig. 1A) were created. PWMs have been widely used to represent peptide binding to HLA molecules and derive the corresponding motifs. As such, they anticipated that naturally processed HLA ligands that come from up to six different HLA molecules could be ideally modeled with a mixture of PWMs. In this probabilistic framework, a total log-likelihood function is defined as follows:
log
P
X
M
,
w
P
p
r
i
o
r
M
=
∑
n
=
1
N
l
o
g
{
∑
k
=
1
K
w
k
∏
l
=
1
L
M
X
l
n
,
l
k
}
+
l
o
g
{
P
p
r
i
o
r
M
}
, where
X
represents the set of peptide ligands,
L
is the length of the peptides,
N
the number of ligands, and
K
the number of motifs.
M
k
is the PWM representing the
k
th motif (with
M
X
l
n
,
l
k
the PWM entry corresponding to the
l
th residue in peptide
X
n
), and
w
k
is the contribution of each motif in the total likelihood (i.e., wherein screening the plurality of library peptides includes determining log-likelihoods of the plurality of library peptides under a weighted position of the binding motif) (Pg. 2493, Col. 1, Para. 3-4; and Pg. 2495, Fig. 1).
Therefore, regarding claims 6 and 16, it would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to modify the method of using reinforcement learning to generate novel protein designs of Skwark et al. in view of Chen et al. with the log-likelihood calculation of Bassani-Sternberg et al. because the method of Bassani-Sternberg provides a method to analyze how binding motifs change with peptide length and predict new cooperative effects between distant residues in HLA-B ligands (Abstract). One of ordinary skill in the art would be able to combine the teachings of Skwark et al. in view of Chen et al. with Bassani-Sternberg et al. with reasonable expectation of success due to the same nature of the problem to be solved, since both incorporate a method for analyzing peptide binding motifs. Therefore, regarding claims 6 and 16, the instant invention is prima facie obvious (MPEP § 2142).
6. Claims 9 and 19 are rejected under 35 U.S.C. 103 as being unpatentable over Skwark et al. in view of Chen et al., as applied to claims 1-2, 7-8, 11-12, and 17-18 above, and further in view of Venkatesh et al. (MHCAttnNet: predicting MHC-peptide bindings for MHC alleles classes I and II using an attention-based deep neural model. Bioinformatics. 36(Supplement_1): i399-i406 (2020); published 07/13/2020).
Skwark et al. in view of Chen et al., as applied to claims 1-2, 7-8, 11-12, and 17-18 above, does not teach comparing the screened plurality of library peptides to a candidate vaccine peptide to determine how the candidate vaccine peptide binds to the MHC.
Regarding claims 9 and 19, Venkatesh et al. teaches that accurate prediction of binding between a major histocompatibility complex (MHC) allele and a peptide plays a major role in the synthesis of personalized cancer vaccines (Abstract). Venkatesh et al. further teaches that personalized cancer vaccines have shown promising results in their early stages. To synthesize a personalized cancer vaccine, first the genomes of cancerous cells are collected, which helps in the identification of tumor-specific peptides called neoepitopes. Neoepitopes, when combined with adjuvants or other immune-stimulatory agents and injected into the patient, helps the immune system identify cancerous cells and kill them using the body’s own T-cells. This way, the human body learns to kill the cancerous cells on its own, without having the risk of autoimmune diseases. For the identification of neoepitopes, next-generation sequencing data from tumor and healthy cells are compared with that of the human reference genome. RNA sequencing narrows the focus to mutations of expressed genes. The potential sequences are validated by using computational models that predict the binding affinity of neoepitopes with the individual’s MHC proteins that would present the neoepitopes to the surface. This filters the candidate neoepitopes for personalized vaccines, as shown in Figure 1 (i.e., comparing the screened plurality of library peptides to a candidate vaccine peptide to determine how the candidate vaccine binds to the MHC) (Pg. i400, Col. 1, Para. 5-6 and Pg. i400, Fig. 1).
Therefore, regarding claims 9 and 19, it would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to modify the method of using reinforcement learning to generate novel protein designs for MHC binding motifs of Swartz et al. in view of Chen et al. with the method of screening candidate vaccine peptides of Venkatesh et al. because Chen et al. discloses that identifying peptides that can bind to MHC class-I molecules plays a vital role in the design of peptide vaccines (Chen et al. Abstract). Venkatesh et al. also discloses a method of screening neoepitopes from MHC proteins to filter candidates for personalized vaccines (Venkatesh et al., Pg. i400, Col. 1, Para. 5-6). One of ordinary skill in the art would be able to combine the teachings of Skwark et al. in view of Chen et al. with Venkatesh et al. with reasonable expectation of success due to the same nature of the problem to be solved, since incorporate a method for determining peptide candidates for MHC proteins for vaccine design. Therefore, regarding claims 9 and 19, the instant invention is prima facie obvious (MPEP § 2142).
7. Claims 10 and 20 are rejected under 35 U.S.C. 103 as being unpatentable over Skwark et al. in view of Chen et al. and Venkatesh et al., as applied to claims 9 and 19 above, further in view of Sekelja et al. (WIPO Publication WO 2020/065023 A1; published 04/02/2020).
Skwark et al. in view of Chen et al. and Venkatesh et al., as applied to claims 9 and 19 above, does not teach creating a vaccine based on the candidate vaccine peptide and administering the vaccine to prevent an illness.
Regarding claims 10 and 20, Sekelja et al. teaches a method for selecting neoepitopes for an individual, by selecting MHC I and/or MHC II binding neoepitopes and ranking them (Abstract). Sekelja et al. further teaches that 10 different neoepitopes (pep1-10) all predicted to bind MCH class I (CDS+ T cell response) have been investigated to determine if they could induce CDS+ T cell responses in a mouse B16-F10 melanoma tumor model. The responses induced by 6 neoepitopes when administered as vaccibody as described herein ("VB10.NEO") are shown in the upper panel of Fig. 6. The CD4 and CDS response when administered as peptide plus poly ICLC adjuvant, RNA and vaccibody are summarized in the lower panel, where white indicates no response, light grey indicates a weak response, medium grey a medium response and dark grey a strong response (i.e., creating a vaccine based on the candidate vaccine peptide and administering the vaccine to prevent an illness) (Pg. 7, Lines 1-10 and Pg. 122, Fig. 6).
Therefore, regarding claims 10 and 20, it would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to modify the method of using reinforcement learning to generate novel protein designs for MHC binding motifs of Swartz et al. in view of Chen et al. and Venkatesh et al. with the vaccine creation of Sekelja et al. because Venkatesh et al. discloses a method of screening neoepitopes from MHC proteins to filter candidates for personalized vaccines (Venkatesh et al., Pg. i400, Col. 1, Para. 5-6), which can be created by the method of Sekelja et al. to improve personalized cancer therapy (Sekelja et al., Pg. 2, Lines 15-23). One of ordinary skill in the art would be able to combine the teachings of Swartz et al. in view of Chen et al. and Venkatesh et al. with Sekelja et al. with reasonable expectation of success due to the same nature of the problem to be solved, since both incorporate a method for screening binding peptides/neoepitopes for MHC proteins. Therefore, regarding claims 10 and 20, the instant invention is prima facie obvious (MPEP § 2142).
Conclusion
No claims allowed.
Inquiries
Any inquiry concerning this communication or earlier communications from the examiner should be directed to DIANA P SANFORD whose telephone number is (571)272-6504. The examiner can normally be reached Mon-Fri 8am-5pm EST.
Examiner interviews are available via telephone, in-person, and video conferencing using a USPTO supplied web-based collaboration tool. To schedule an interview, applicant is encouraged to use the USPTO Automated Interview Request (AIR) at http://www.uspto.gov/interviewpractice.
If attempts to reach the examiner by telephone are unsuccessful, the examiner’s supervisor, Karlheinz Skowronek can be reached at (571)272-9047. The fax phone number for the organization where this application or proceeding is assigned is 571-273-8300.
Information regarding the status of published or unpublished applications may be obtained from Patent Center. Unpublished application information in Patent Center is available to registered users. To file and manage patent submissions in Patent Center, visit: https://patentcenter.uspto.gov. Visit https://www.uspto.gov/patents/apply/patent-center for more information about Patent Center and https://www.uspto.gov/patents/docx for information about filing in DOCX format. For additional questions, contact the Electronic Business Center (EBC) at 866-217-9197 (toll-free). If you would like assistance from a USPTO Customer Service Representative, call 800-786-9199 (IN USA OR CANADA) or 571-272-1000.
/D.P.S./Examiner, Art Unit 1687
/Lori A. Clow/Primary Examiner, Art Unit 1687
Prosecution Insights