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 .
Election/Restrictions
Applicant's election with traverse of species A, training data including MHC molecules and the MIL classifier is trained to predict a peptide sequence presented by any of the MHC molecules (claims 1-11 and 13) in the reply filed on 17 Aug. 2026 is acknowledged. The traversal is on the ground(s) that the ISA did not reject the claims for lack of unity, and Applicant submits that the present case falls into the situation where the MPEP urges the Examiner to take the broad practical approach and Examine the claims instead of making a narrow academic distinction (Applicant’s remarks at pg. 2, para. 2). This is not found persuasive for the following reasons. The species lack unity of invention for the reasons set forth in the restriction requirement mailed 17 July 2026, and Applicant has not provided any arguments that the species do not lack unity of invention.
The requirement is still deemed proper and is therefore made FINAL.
Claims 14-15 are withdrawn from further consideration pursuant to 37 CFR 1.142(b), as being drawn to a nonelected species, there being no allowable generic or linking claim. Applicant timely traversed the restriction (election) requirement in the reply filed on 17 Aug. 2026.
Status of Claims
Claims 1-15 are pending.
Claims 14-15 are withdrawn from further consideration pursuant to 37 CFR 1.142(b), as being drawn to a nonelected species, there being no allowable generic or linking claim.
Claims 1-13 are rejected.
Claim 1 is objected to.
Priority
Applicant’s claim for the benefit of a prior-filed application, PCT/EP2021/056387 filed 12 March 2021, under 35 U.S.C. 119(e) or under 35 U.S.C. 120, 121, 365(c), or 386(c) is acknowledged.
Acknowledgment is made of applicant’s claim for foreign priority to EP20201557.4 filed 13 Oct. 2020 under 35 U.S.C. 119 (a)-(d). Receipt is acknowledged of certified copies of papers required by 37 CFR 1.55.
Therefore, the effective filing date of the claimed invention is 13 Oct. 2020.
Information Disclosure Statement
The information disclosure statement(s) (IDS) submitted on 07 July 2023 is in compliance with the provisions of 37 CFR 1.97. Accordingly, the list of cited references was considered in full by the examiner.
Drawings
The drawings filed 11 April 2023 are objected to because:
the drawings fail to comply with 37 CFR 1.84(u)(1), which states view numbers must be preceded by the abbreviation "FIG.". The view numbers should be relabeled as “FIG. 1”, “FIG. 2”, etc.
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. The figure or figure number of an amended drawing should not be labeled as “amended.” If a drawing figure is to be canceled, the appropriate figure must be removed from the replacement sheet, and where necessary, the remaining figures must be renumbered and appropriate changes made to the brief description of the several views of the drawings for consistency. Additional replacement sheets may be necessary to show the renumbering of the remaining figures. 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.
Specification
The replacement abstract filed 11 April 2023 has been entered.
Applicant is reminded of the proper language and format for an abstract of the disclosure.
The abstract should be in narrative form and generally limited to a single paragraph on a separate sheet within the range of 50 to 150 words in length. The abstract should describe the disclosure sufficiently to assist readers in deciding whether there is a need for consulting the full patent text for details.
The language should be clear and concise and should not repeat information given in the title. It should avoid using phrases which can be implied, such as, “The disclosure concerns,” “The disclosure defined by this invention,” “The disclosure describes,” etc. In addition, the form and legal phraseology often used in patent claims, such as “means” and “said,” should be avoided.
The abstract of the disclosure is objected to because:
the abstract is more than 150 words.
A corrected abstract of the disclosure is required and must be presented on a separate sheet, apart from any other text. See MPEP § 608.01(b).
Claim Objections
Claim 1 is objected to because of the following informalities:
Claim 1 recites “wherein the training data ar organized…” in the first limitation, which is a typographical error and should recite “are organized”.
Appropriate correction is required.
Claim Interpretation
Claim 1 recites “A computer-implemented method for predicting binding and presentation of peptides by major histocompatibility complex (MHC) molecules, the method comprising….predicting the label of new instances…”. MPEP 2111.02 II. states if the body of a claim fully and intrinsically sets forth all of the limitations of the claimed invention, and the preamble merely states, for example, the purpose or intended use of the invention, rather than any distinct definition of any of the claimed invention’s limitations, then the preamble is not considered a limitation and is of no significance to claim construction. Shoes by Firebug LLC v. Stride Rite Children’s Grp., LLC, 962 F.3d 1362, 2020 USPQ2d 10701 (Fed. Cir. 2020). In the instant case, the preamble of claim 1 recites an intended use because the body of the claim sets forth all of the limitations in the claim, but does not require the prediction is a prediction of binding and presentation of peptides by MHC molecules.
Claim Rejections - 35 USC § 112(a)
The following is a quotation of the first paragraph of 35 U.S.C. 112(a):
(a) IN GENERAL.—The specification shall contain a written description of the invention, and of the manner and process of making and using it, in such full, clear, concise, and exact terms as to enable any person skilled in the art to which it pertains, or with which it is most nearly connected, to make and use the same, and shall set forth the best mode contemplated by the inventor or joint inventor of carrying out the invention.
Claims 1-11 are rejected under 35 U.S.C. 112(a) as failing to comply with the written description requirement. The claim(s) contains subject matter which was not described in the specification in such a way as to reasonably convey to one skilled in the relevant art that the inventor or a joint inventor, at the time the application was filed, had possession of the claimed invention.
Claims 1 and 10, and claims dependent therefrom, recite “A computer-implemented method for predicting binding and presentation of peptides…collecting or generating training data, wherein the training data includes a set of MHC molecules present in a biological sample as well as a set of observed peptide sequences that are presented by at least one of the MHC molecules present in the biological sample, wherein it is not known to which specific of the MHC molecules a peptide sequence is bound, and wherein the training data are organized in bags with each bag having a set of training instances, wherein labels are known for the bags, but unknown for the training instances ….; predicting the label of new instances by applying the MIL classifier fθ directly and/or predicting the label of new bags by applying the MIL classifier fθ to each instance of a respective bag and aggregating the results among all instances of the respective bag.
Claim 11 recites essentially the same limitations above as claim 1.
As presently recited, claims encompass using any labels and using the MIL to make a prediction of any label, using training data including MHC molecules and observed peptide sequences presented by at least one of the MHC molecules. MPEP 2163 II A. 3(a) II states the written description requirement for a claimed genus may be satisfied through sufficient description of a representative number of species by actual reduction to practice (see i)(A) above), reduction to drawings (see i)(B) above), or by disclosure of relevant, identifying characteristics, i.e., structure or other physical and/or chemical properties, by functional characteristics coupled with a known or disclosed correlation between function and structure, or by a combination of such identifying characteristics, sufficient to show the inventor was in possession of the claimed genus (see i)(C) above). In the instant case, the specification does not provide support for the genus of labels and predictions encompassed by the claims, using training data including MHC molecules and observed peptide sequences presented by at least one of the MHC molecules as claimed. For example, Applicant’s specification at para. [0059] and claims 14-15 discloses various embodiments in which an MIL classifier is trained to predict a cancerous image in a region or a topic of documents, which is only disclosed as being performed using training data relating to stained images and text documents. Applicant’s specification does not disclose how these labels/predictions are performed using the training set with MHC molecules and observed peptide sequences. Therefore, Applicant’s specification does not provide support for the claimed genus of making a prediction for any label as claimed.
For the reasons discussed above, the specification does not provide a sufficient disclosure of the limitations above recited in claims 1-11 to demonstrate to one of ordinary skill in the art that the inventor possessed the invention at the time the application was filed. For more information regarding the written description requirement, see MPEP §2161.01- §2163.07(b).
It is noted that the preamble of claim 1 recites the intended use of predicting binding and presentation of peptides by major histocompatibility complex (MHC) molecules. To overcome the rejection, it is suggested Applicant amend claim 1 to make clear the labels and prediction relate to predicting binding and presentation of peptides by MHC molecules.
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.
Claims 1-13 are rejected under 35 U.S.C. 112(b) as being indefinite for failing to particularly point out and distinctly claim the subject matter which the inventor or a joint inventor, or for pre-AIA the applicant regards as the invention.
Claims 1 and 11-12, and claims dependent therefrom, are indefinite for recitation of “predicting the label of new instances…and/or predicting the label of new bags…”. There is insufficient antecedent basis for a label of new instances and a label of new bags, because the claim previously does not recite a label for new instances or new bags (i.e. one label for multiple bags/instances), and furthermore, claims 1 and 11-12 previously recite “wherein labels are known for the bags…” (i.e. multiple labels), which encompasses one bag having multiple labels or each bag having different labels. As a result, it is further unclear which label is being referenced. For purpose of examination, the limitations are interpreted to mean “predicting a label”.
Claims 4 and 12, and claims dependent therefrom, are indefinite for recitation of “wherein individual training instances from positively labeled bags are weighted by…”. Examples of claim language, although not exhaustive, that may raise a question as to the limiting effect of the language in a claim are: (A) "adapted to" or "adapted for" clauses; (B) "wherein" clauses; and (C) "whereby" clauses. In the instant case, claim 1, from which claim 4 depends, recites “wherein the training data are organized in bags with each bag…, wherein labels are known for the bags”. However, claim 1 and claim 12 do not recite what these labels are, such as positively or negatively labelled. As a result, it is unclear if the limitation of claims 4 and 12 of the training instances being weighted is intended to be contingent upon any individual training instances being from positively labeled bags, or if claims 4 and 12 intend to further limit the labels of the bags to include positively labeled bags containing weighted training instances. For purpose of examination, the limitation is interpreted to be contingent upon the presence of positively labeled bags.
Claims 4 and 12, and claims dependent therefrom, are indefinite for recitation of a “calibrated current model confidence function”. The metes and bounds of what confidence functions fall within the scope of a calibrated current model confidence function are not clear. Specifically, it is unclear in what way a calibrated model confidence function is intended to be different than a calibrated current model confidence function. A review of Applicant’s specification only repeats the language in the claims and does not serve to clarify the metes and bounds of the term. Clarification is requested via claim amendment.
Claim 7 is indefinite for recitation of “the parameters of the MIL classifier”. There is insufficient antecedent basis for this limitation in the claim because claims 1 and 5, from which claim 7 depends, do not recite any parameters of the MIL classifier. Furthermore, given a MIL classifier may only have one parameter, the MIL classifier itself does not provide antecedent basis for “the parameters”. For purpose of examination, the limitation is interpreted to mean “training parameters of the MIL classifier…”.
Claim 7 is indefinite for recitation of “the Loss function L(θ) that includes a probability calibration function C…”. There is insufficient antecedent basis for this limitation in the claim because claims 1 and 5, from which claim 7 depends, do not recite a loss function L(θ) including a probability calibration function. Instead, claim 1 recites “a loss function”.
Claim 7 is indefinite for recitation of “the probabilities pi of yi...”. There is insufficient antecedent basis for “the probabilities…” because claims 1 and 5 do not previously recite any probabilities. For purpose of examination, the limitation is interpreted to mean “configured to predict….probabilities…”.
Claim 7 is indefinite for recitation of “the tuple (si, ai), where si is the peptide and ai is the jth MHC molecule in Ai”. Claim 5, from which claim 7 depends, recites “si is a peptide sequence” and claim 1, from which claim 7 depends, recites “a set of observed peptide sequences”. Therefore, it is not clear which peptide “the peptide” is referring to. Furthermore, given “ai” has a subscript “i" and claim 5 does not mention a subscript “j”, it is further unclear which MHC molecule “the jth MHC molecule in Ai” is referring to.
Claim 9 is indefinite for recitation of “the peptides with the highest likelihood of being presented…”. There is insufficient antecedent basis for “the peptides with the highest likelihood of being presented”, because the claim does not previously recite any likelihood of being presented, and furthermore, the MIL classifier is not recited to predict a likelihood of being presented. Therefore, it is not clear what peptides, “the peptides with the highest likelihood” are referring to.
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 1-13 are rejected under 35 U.S.C. 101 because the claimed invention is directed to one or more judicial exceptions without significantly more.
The Supreme Court has established a two-step framework for this analysis, wherein a claim does not satisfy § 101 if (1) it is “directed to” a patent-ineligible concept, i.e., a law of nature, natural phenomenon, or abstract idea, and (2), if so, the particular elements of the claim, considered “both individually and as an ordered combination,” do not add enough to “transform the nature of the claim into a patent-eligible application.” Elec. Power Grp., LLC v. Alstom S.A., 830 F.3d 1350, 1353 (Fed. Cir. 2016) (quoting Alice, 134 S. Ct. at 2355). Applicant is also directed to MPEP 2106.
Step 1: The instantly claimed invention (claims 1 and 10-12 being representative) is directed a method, product, and system for predicting binding and presentation of peptides by major histocompatibility complex (MHC) molecules. Therefore, the instantly claimed invention falls into one of the four statutory categories. [Step 1: YES]
Step 2A: First it is determined in Prong One whether a claim recites a judicial exception, and if so, then it is determined in in Prong Two if the recited judicial exception is integrated into a practical application of that exception.
Step 2A, Prong 1: Under the MPEP § 2106.04, the Step 2A (Prong 1) analysis requires determining whether a claim recites an abstract idea, law of nature, or natural phenomenon.
Claims 1 and 10-11 recite the following steps which fall under the mathematical concepts, mental processes, and/or certain methods of organizing human activity groupings of abstract ideas:
collecting or generating training data, wherein the training data includes a set of MHC molecules present in a biological sample as well as a set of observed peptide sequences that are presented by at least one of the MHC molecules present in the biological sample, wherein it is not known to which specific of the MHC molecules a peptide sequence is bound, and
wherein the training data are organized in bags with each bag having a set of training instances, wherein labels are known for the bags, but unknown for the training instances (claims 1 and 10) or organizing the training data in bags with each bag having a set of training instances (claim 11);
using a loss function to train a multiple instance learning (MIL) classifier fθ at an instance-level; and
predicting the label of new instances by applying the MIL classifier fθ directly and/or predicting the label of new bags by applying the MIL classifier fθ to each instance of a respective bag and aggregating the results among all instances of the respective bag.
Claim 12 recite the following steps which fall under the mathematical concepts, mental processes, and/or certain methods of organizing human activity groupings of abstract ideas:
collecting or generating training data, wherein the training data includes bags with each bag having a set of training instances, wherein labels are known for the bags, but unknown for the training instances;
training an MIL classifier at an instance-level by using a loss function that explicitly accounts for a model confidence in the model predictions during training, wherein individual training instances from positively labelled bags are weighted by a calibrated current model confidence function; and
predicting the label of new instances by applying the MIL classifier directly and/or predicting the label of new bags by applying the MIL classifier to each instances of a respective bag and aggregating the results among all instances of the respective bag.
The identified claim limitations falls into one of the groups of abstract ideas of mathematical concepts and/or mental processes for the following reasons. In this case, collecting or generating training data organized into bags can be practically performed in the mind by organizing information relating to MHC molecules present in a sample into groups (i.e. bags). The step of training a MIL classifier using a loss function at an instance level recites the mathematical concept of a mathematical calculation (i.e. iteratively calculating a loss). Predicting a label of new instances using the trained MIL classifier and aggregating the results among instances of the respective bag also recites a mathematical concept because the limitation encompasses applying a trained MIL regression classifier to calculate a label, and then aggregating the results (e.g. performing addition, taking an average). The predicting limitation also recites a mental process because the human mind is equipped to input values into a trained classifier, carry out mathematical operations (e.g. addition, multiplication), and then taking an average. Further regarding the limitations reciting a mental process, other than reciting these limitations are carried out by a computer, nothing in the claims precludes the steps from being practically performed in the mind. See MPEP 2106.04(a)(2) I. and III.
Dependent claims 2-9 and 13 further recite an abstract idea and/or are part of the abstract idea of claims 1 and 12 above. Dependent claims 2-3 further limit the mathematical concept of training the MIL classifier of claim 1 and claim 2 further recites a mathematical equation. Dependent claims 4-5 further limit the mental process of collecting training data to involve weighting the training data and providing the training data in the form of a set of triples. Dependent claim 4 further recites the mathematical concept of weighting individual training instances using a calibrated current model confidence function. Dependent claim 6 further limits the mental process of collecting training data to obtain the training data from mass spectrometry experiments, which encompasses analyzing and organizing data form mass spectrometry experiments into a training set. Dependent claim 7 further limits the mathematical concept of training parameters of the MIL classifier by the loss function to include a probability calibration function for predicting probabilities. Dependent claim 8 further recites the mental process of providing a peptide sequence and a set of MHC molecules as input to the MIL classifier, and the mental process and mathematical concept of applying the MIL classifier to predict whether the peptide sequence will be presented by any of the MHC molecules. Dependent claim 9 further recites the mental process and mathematical concept of using the MIL classifier to make predictions for all combinations of peptide sequences and MHC molecules, and the mental process of determining peptides with the highest likelihood of being presented as candidates for a vaccine. Dependent claim 13 further limits the mental process of collecting training data to collect training data including a set of MHC molecules and observed peptide sequences, and further limits the mathematical concept of training the MIL classifier to make certain predictions. Therefore, claims 1-13 recite an abstract idea. [Step 2A, Prong 1: YES]
Step 2A: Prong 2: Under the MPEP § 2106.04, the Step 2A, Prong 2 analysis requires identifying whether there are any additional elements recited in the claim beyond the judicial exception(s), and evaluating those additional elements to determine whether they integrate the exception into a practical application of the exception. This judicial exception is not integrated into a practical application for the following reasons.
Claims 2-9 and 13 do not recite any elements in addition to the judicial exception, and therefore are part of the judicial exception.
The additional elements of claims 1 and 10-12 include:
a computer (claims 1 and 12);
a tangible, non-transitory computer-readable medium (claim 10); and
a system comprising one or more processors (claim 11).
The additional elements of a computer, non-transitory computer-readable medium, and system comprising a processor are generic computer components that are merely used to carry out the abstract idea identified above. The courts have found the use of a computer or other machinery in its ordinary capacity for economic or other tasks (e.g., to receive, store, or transmit data) or simply adding a general purpose computer or computer components after the fact to an abstract idea (e.g., a fundamental economic practice or mathematical equation) does not integrate a judicial exception into a practical application. See MPEP 2106.05(f).
Therefore, the additionally recited elements amount to mere instructions to apply the exception by a computer and, as such, the claims as a whole do not integrate the abstract idea into practical application. Thus, claims 1-13 are directed to an abstract idea. [Step 2A, Prong 2: NO]
Step 2B: In the second step it is determined whether the claimed subject matter includes additional elements that amount to significantly more than the judicial exception. See MPEP § 2106.05.
The claims do not include any additional steps appended to the judicial exception that are sufficient to amount to significantly more than the judicial exception.
Claims 2-9 and 13 do not recite any elements in addition to the judicial exception, and therefore are part of the judicial exception.
The additional elements of claims 1 and 10-12 include:
a computer (claims 1 and 12);
a tangible, non-transitory computer-readable medium (claim 10); and
a system comprising one or more processors (claim 11).
The additional elements of a computer, non-transitory computer-readable medium, and system comprising a processor are conventional computer components that are merely used to carry out the abstract idea identified above. The courts have found the use of a computer or other machinery in its ordinary capacity for economic or other tasks (e.g., to receive, store, or transmit data) or simply adding a general purpose computer or computer components after the fact to an abstract idea (e.g., a fundamental economic practice or mathematical equation) does not provide significantly more. See Affinity Labs v. DirecTV, 838 F.3d 1253, 1262, 120 USPQ2d 1201, 1207 (Fed. Cir. 2016) (cellular telephone); TLI Communications LLC v. AV Auto, LLC, 823 F.3d 607, 613, 118 USPQ2d 1744, 1748 (Fed. Cir. 2016) (computer server and telephone unit).
Therefore, taken alone, the additional elements do not amount to significantly more than the above-identified judicial exception(s). Even when viewed as a combination, the additional elements fail to transform the exception into a patent-eligible application of that exception. Thus, the claims as a whole do not amount to significantly more than the exception itself. [Step 2B: NO]
Therefore, the instantly rejected claims are not drawn to eligible subject matter as they are directed to an abstract idea (and/or natural correlation) without significantly more. For additional guidance, applicant is directed generally to the MPEP § 2106.
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.
This application currently names joint inventors. In considering patentability of the claims the examiner presumes that the subject matter of the various claims was commonly owned as of the effective filing date of the claimed invention(s) absent any evidence to the contrary. Applicant is advised of the obligation under 37 CFR 1.56 to point out the inventor and effective filing dates of each claim that was not commonly owned as of the effective filing date of the later invention in order for the examiner to consider the applicability of 35 U.S.C. 102(b)(2)(C) for any potential 35 U.S.C. 102(a)(2) prior art against the later invention.
Claims 1, 3-6, and 9-11 are rejected under 35 U.S.C. 103 as being unpatentable over EL-Manzalawy in view of Park (2020) and Reynisson (2020), as evidenced by Shevade (2000).
Cited references:
EL-Manzalawy, Predicting MHC-II binding affinity using multiple instance regression, 2011, IEE?ACM Trans Comput Biol Bioinform, 8(4), pg. 1067-1079 (cited in IDS filed 07 July 2023)
Park et al., Bayesian multiple instance regression for modeling immunogenic neoantigens, 2020 13 May, Statistical Methods in Medical Research, 29(10), pg. 3032-3047 (cited in IDS filed 07 July 2023; included in PTO-892 as cited pg. numbers do not correspond to copy in IDS);
Reynisson et al., Improved prediction of MHC II antigen presentation through integration and motif deconvolution of mass spectrometry MHC eluted ligand data, Feb. 2020, bioRxiv, pg. 1-34;
Shevade, Improvements to the SMO Algorithm for SVM Regression, 2000, IEE Transactions on Neural Networks, 11(5), pg. 1188-1193.
Regarding claims 1 and 11, EL-Manzalawy discloses a method for predicting MHC binding affinity using multiple instance learning (Abstract), wherein the method comprises the following steps:
EL-Manzalawy discloses collecting training data including a set of MHC-II molecules and experimentally determined binding affinities of MHC-II peptides to the MHC-II molecules (pg. 6, para. 2-4; Tables 1-3). EL-Manzalawy discloses the training data is organized into bags with all 9-mer subsequences of a MHC-II peptide sequence in each bag (pg. 5, para. 4-5), wherein each bag has a set of training instances (pg. 4, para. 3, e.g. bags of instances), and wherein labels are associated with the bags of 9-mers and not individual 9-mers (i.e. labels are known for bags, but unknown for training instances) (pg. 5, para. 4-5, e.g. organization of peptide data into bags).
EL-Manzalawy discloses training a multiple instance learning (MIL) classifier, wherein the MIL classifier is trained at a meta-instance level (pg. 7, para. 1). EL-Manzalawy discloses the MIL classifier is a support vector regression classifier, citing reference [54] (Shevade). The MIL classifier of EL-Manzalawy is trained using a loss function, as evidenced by Shevade, which discloses the SVM regression model (used in EL-Manzalawy) uses a loss function (pg. 1188, col. 2, para. 2-3).
EL-Manzalawy discloses predicting a binder or non-binder label of new bags by applying the MIL classifier to a bag of 9-mers (i.e. each instance of a respective bag) representing a query peptide (pg. 6, para. 4; pg. 7, para. 4 and 6).
Regarding claims 1, 6, and 10-11, EL-Manzalawy does not disclose the following limitations:
Further regarding claim 1¸ EL-Manzalawy does not disclose the training of the MIL is at an instance-level, or that the predicting comprises predicting the label of new instances by applying the MIL classifier direction, or that the label of new bags comprising applying the MIL classifier to each instance of a respective bag and aggregation the results among all instances of the respective bag.
However, Park discloses a method of performing multiple instance regression for modeling peptides represented by MHC proteins (e.g. neoantigens) (Abstract; pg. 3033, para. 4), and discloses an instance-based type of multiple instance learning (pg. 3033, para. 2-3). Park discloses training a regression model at an instance level to make instant level predictions (pg. 3037, para. 2-5), and further discloses an aggregation function which combines the instance level predictions (pg. 3035, para. 6; pg. 3037, para. 4). Park discloses that meta-instance multiple instance learning methods (as used in EL-Manzalawy), unlike instance-level learning methods, cannot model individual instances or capture their relationship with bag-level responses, which is of great interest in many regression settings (pg. 3033, para. 2-3).
It would have been prima facie obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to have modified the multiple instance MIL of EL-Manzalawy to have used the instance-based multiple instance learning classifier to make instance level predictions and aggregate results of instances of a bag, as shown by Park above. One of ordinary skill in the art would have been motivated to combine the methods of EL-Manzalawy and Park in order to capture the relationship between individual instances and bag level responses, as shown by Park (pg. 3033, para. 2-3). This modification would have had a reasonable expectation of success given both EL-Manzalawy and Park use multiple instance learning to predict peptides interacting with MHC molecules, Park uses attributes of peptides in the instance-level modeling, and therefore the method of Park is applicable to EL-Manzalawy.
Further regarding claims 1, 6, and 11, EL-Manzalawy does not disclose that it is not known to which specific of the MHC molecules a peptide sequence is bound, as recited in claims 1 and 11, or obtaining the training data from mass spectrometry experiments respectively, as recited in claim 6. Instead, EL-Manzalawy discloses the training data is experimentally determined affinities of MHC-II peptides to the MHC-II molecules (pg. 6, para. 2-4; Tables 1-3).
However, Reynisson discloses a method of improved prediction of MHC II antigen presentation through mass spectrometry eluted ligand data (Abstract), which includes training a machine learning model on large-scale eluted MHC II ligand mass spectrometry data to predict whether an antigen is presented by MHC II (i.e. a peptide binding to MHC II) (Abstract; pg. 6, para. 3 to pg. 7, para. 2; pg. 8, para. 3; pg. 9, para. 4, e.g. a pan-specific predictor of MHC II antigen presentation trained on EL data). Reynisson discloses the result of the mass spectrometry assay used in the training is a list of peptide sequences restricted to at least one of the MHC II molecules expressed by the interrogated cell line (i.e. it is not known which MHC molecule a peptide sequence binds in mass-spectrometry data) (pg. 3, para. 2). Reynisson discloses that most prediction methods are trained on either binding affinity (BA) data or MHC MS eluted ligand (EL) data (pg. 4, para. 2), and further discloses that evidence suggests that peptide MHCII binding affinity is a relative weak correlate of MHC antigen presentation, and several studies have demonstrated that MHC-II peptide prediction models can benefit from being trained on immunopeptidome data obtained by liquid chromatography coupled mass spectrometry (pg. 3, para. 2).
It would have been prima facie obvious to one of ordinary skill in the art, before the effective filing date of the claimed invention, to have modified the training data of EL-Manzalawy to have used eluted MHC II ligand mass spectrometry data, as shown by Reynisson above. One of ordinary skill in the art would have been motivated to combine the methods of EL-Manzalawy and Reynisson in order to provide improve performance of MHC-II peptide prediction models, given evidence suggests that peptide MHCII binding affinity is a relative weak correlate of MHC antigen presentation, as shown by Reynisson (pg. 3, para. 2). This modification would have had a reasonable expectation of success given Reynisson discloses binding affinity data (as used in EL-Manzalawy) or MS data are commonly used to make the same predictions of a peptide binding to an MHC-II molecule, and thus the data of Reynisson is applicable to the multiple instance learning method of EL-Manzalawy.
Regarding claims 10-11, EL-Manzalawy does not explicitly disclose the method is stored on a non-transitory computer readable medium or carried out by a system comprising a processor configured to perform the method. However broadly claiming an automated means to replace a manual function to accomplish the same result does not distinguish over the prior art. See Leapfrog Enters., Inc. v. Fisher-Price, Inc., 485 F.3d 1157, 1161, 82 USPQ2d 1687, 1691 (Fed. Cir. 2007). Furthermore, implementing a known function on a computer has been deemed obvious to one of ordinary skill in the art if the automation of the known function on a general purpose computer is nothing more than the predictable use of prior art elements according to their established functions. KSR Int’l Co. v. Teleflex Inc., 550 U.S. 398, 417, 82 USPQ2d 1385, 1396 (2007);
Regarding the dependent claims:
Regarding claim 3, Shevade discloses the loss function (used in EL-Manzalawy) accounts for model confidence by penalizing error if the error is over a certain threshold (pg. 1188, col. 2, para. 4).
Regarding claim 4, the limitation regarding training instances from positively labeled bags being weighted is a contingent limitation, contingent upon there being positively labeled bags (see 112(b) interpretation above). However, because the claim does not require positively labeled bags, the limitation of claim 4 is not required by the claim. See MPEP 2111.04 II.
Regarding claim 5, EL-Manzalawy the training data includes a set of MHC-II molecules and experimentally determined binding affinities of MHC-II peptides to the MHC-II molecules (i.e. a peptide sequence and MHC molecules) (pg. 6, para. 2-4; Tables 1-3). EL-Manzalawy further discloses categorizing the binding affinities of the training set to be either binders or non-binders using a threshold (i.e. a binary label) (pg. 6, para. 2). While EL-Manzalawy does not explicitly disclose the above training data is provided in the recited form, the form of the training data is interpreted as a matter of design choice, and Applicant has not disclosed that this feature provides an advantage, is used for a particular purpose, or solves a stated problem when compared to the form of the training data in EL-Manzalawy. For example, there is no functional link between this recited form of the training data and how the machine learning model is trained using the training data. Therefore, the form of training data of EL-Manzalawy would perform equally as well in training a MIL model, and such a modification fails to patentably distinguish over EL-Manzalawy.
Regarding claim 9, EL-Manzalawy discloses using the MIL model to make predictions for the combinations of peptide sequences and MHC molecules in a validation set (pg. 8, para. 2-3; Table II). EL-Manzalawy further discloses identifying the highest scoring 9-mer of a set of peptides (pg. 10, para. 1), and discloses the identification of such peptides are important for designing vaccines (pg. 1, para. 1 to pg. 2, para. 2), demonstrating peptides determined as having a high likelihood are vaccine candidates.
Therefore the invention is prima facie obvious.
Claim 2 is rejected under 35 U.S.C. 103 as being unpatentable over El-Manzalawy in view of Park and Reynisson, as applied to claim 1 above, and further in view of Gildenblat (2020)
Cited reference: Gildenblat et al., Certainty Pooling for Multiple Instance Learning, Aug. 2020, arXiv, pg. 1-13.
Regarding claim 2, El-Manzalawy in view of Park and Reynisson disclose the method of claim 1, as applied above.
PNG
media_image1.png
61
398
media_image1.png
Greyscale
Further regarding claim 2, El-Manzalawy in view of Park and Reynisson, as applied to claim 1 above, do not disclose the loss function is:
wherein xi1,xiz,…,xim are the instances of bag i, yi are the associated bag labels, h is a permutation-invariant pooling function, N is the number of bags and M is the number of instances in each bag.
However, Gildenblat discloses pooling functions for multiple instance learning (Abstract), and discloses that bag level predictions are derived from the multiple instances through application of a permutation invariant pooling operator on instance predictions or embeddings (Abstract). Gildenblat discloses, that common pooling operators include a mean-pooling equation that predicts the bag level prediction Zm from the instance predictions, Zm = 1/K∑ Kk=1 hkm. Accordingly, the loss calculated according to El-Manzalawy in view of Park and Reynisson between such bag-level predictions and known bag labels determined by an invariant pooling function read on the above loss function.
It would have been prima facie obvious to one of ordinary skill in the art, before the effective filing date of the claimed invention, to have modified the loss function of El-Manzalawy in view of Park and Reynisson, as applied to claim 1 above, to have used the mean-pooling equation of Gildenblat disclosed above. One of ordinary skill in the art would have been motivated to combine the methods of El-Manzalawy in view of Park and Reynisson with Gildenblat based on the simple substitution of the pooling function of El-Manzalawy in view of Park and Reynisson (Park: pg. 3037, para. 3, e.g. aggregation function) with the mean-pooling function of Gildenblat. One of ordinary skill in the art could have substituted the pooling function of El-Manzalawy in view of Park and Reynisson with the pooling function of Gildenblat, and the results of the substitution would have predictable resulted in the bag-level predictions being determined according to the pooling function of Gildenblat, and subsequently used in the loss function of El-Manzalawy in view of Park and Reynisson, thus arriving at the invention of claim 2.
Therefore the invention is prima facie obvious.
Claim 8 is rejected under 35 U.S.C. 103 as being unpatentable over El-Manzalawy in view of Park and Reynisson, as applied to claim 1 above, and further in view of Alvarez (2019).
Cited reference: Alvarez et al., NNAlign_MA; MHC Peptidome Deconvolution for Accurate MHC Binding Motif Characterization and Improved T-cell Epitope Predictions, 2019, Mol Cell Proteomics, 18(12), pg. 2459-2477.
Regarding claim 8, El-Manzalawy in view of Park and Reynisson disclose the method of claim 1, as applied above.
Further regarding claim 8, El-Manzalawy discloses evaluating the predictive performance of the MIL model on validation sets (pg. 8, para. 5 to pg. 9, para. 1), which comprises providing a query peptide I the form of a bag of 9-mers as input to the model (pg. 6, para. 4).
El-Manzalawy discloses the model then predicts the affinity of the query peptide for the MHC molecule (pg. 6, para. 4; Table II).
Further regarding claim 8, El-Manzalawy in view of Park and Reynisson, as applied to claim 1 above, do not disclose the MHC molecules are provided as input into the MIL model.
However, Alvarez discloses a method for predicting MHC binding (Abstract), which includes labeling Mass spectrometry eluted ligand (MS EL) data (i.e. peptides associated with MHC molecules, as in El-Manzalaway in view of Park and Reynisson) with its corresponding MHC allele for use in a machine learning model to predict MHC binding (pg. 2460, col. 2, para. 2). Alvarez discloses the poly-specific nature of MS-EL libraries constitutes a challenge in terms of data analysis and interpretation, where to learn specific MHC rules for antigen presentation, one must first associate each ligand to its presenting MHC molecules (pg. 2460, col. 1, para. 3-4).
It would have been prima facie obvious to one of ordinary skill in the art, before the effective filing date of the claimed invention, to have modified the method of El-Manzalawy in view of Park and Reynisson, as applied to claim 1 above, to have labeled the input peptides with MHC allele information, as shown by Alvarez above. One of ordinary skill in the art would have been motivated to combine the methods of El-Manzalawy in view of Park and Reynisson with Alvarez, in order to learn specific MHC rules for antigen presentation, as shown by Alvarez (pg. 2460, col. 1, para. 3-4). This modification would have had a reasonable expectation of success given both El-Manzalawy in view of Park and Reynisson and Alvarez using machine learning with MS EL data to predict binding of peptides to MHC molecules, and thus the MHC allele labels of Alvarez are applicable to the peptide data of El-Manzalawy in view of Park and Reynisson.
Therefore, the invention is prima facie obvious.
Claim 12 is rejected under 35 U.S.C. 103 as being unpatentable over EL-Manzalawy in view of Park (2020) as evidenced by Shevade (2000).
Cited references:
EL-Manzalawy, Predicting MHC-II binding affinity using multiple instance regression, 2011, IEE?ACM Trans Comput Biol Bioinform, 8(4), pg. 1067-1079;
Park et al., Bayesian multiple instance regression for modeling immunogenic neoantigens, 2020 13 May, Statistical Methods in Medical Research, 29(10), pg. 3032-3047 (cited in IDS filed 07 July 2023);
Reynisson et al., Improved prediction of MHC II antigen presentation through integration and motif deconvolution of mass spectrometry MHC eluted ligand data, Feb. 2020, bioRxiv, pg. 1-34; and
Shevade, Improvements to the SMO Algorithm for SVM Regression, 2000, IEE Transactions on Neural Networks, 11(5), pg. 1188-1193.
Regarding claim 12, EL-Manzalawy discloses a method for predicting MHC binding affinity using multiple instance learning (Abstract), wherein the method comprises the following steps:
EL-Manzalawy discloses collecting training data including a set of MHC-II molecules and experimentally determined binding affinities of MHC-II peptides to the MHC-II molecules (pg. 6, para. 2-4; Tables 1-3). EL-Manzalawy discloses the training data is organized into bags with all 9-mer subsequences of a MHC-II peptide sequence in each bag (pg. 5, para. 4-5), wherein each bag has a set of training instances (pg. 4, para. 3, e.g. bags of instances), and wherein labels are associated with the bags of 9-mers and not individual 9-mers (i.e. labels are known for bags, but unknown for training instances) (pg. 5, para. 4-5, e.g. organization of peptide data into bags). The limitation regarding training instances from positively labeled bags being weighted is a contingent limitation, contingent upon there being positively labeled bags (see 112(b) interpretation above). However, because the claim does not require positively labeled bags, the weighing by a calibrated current model confidence function is not required by the claim. See MPEP 2111.04 II.
EL-Manzalawy discloses training a multiple instance learning (MIL) classifier, wherein the MIL classifier is trained at a meta-instance level (pg. 7, para. 1). EL-Manzalawy discloses the MIL classifier is a support vector regression classifier, citing reference [54] (Shevade). The MIL classifier of EL-Manzalawy is trained using a loss function that explicitly accounts for a model confidence in the model predictions during training, as evidenced by Shevade, which discloses the SVM regression model (used in EL-Manzalawy) uses a loss function that penalizes error if the error is above a threshold (pg. 1188, col. 2, para. 2-3).
EL-Manzalawy discloses predicting a binder or non-binder label of new bags by applying the MIL classifier to a bag of 9-mers (i.e. each instance of a respective bag) representing a query peptide (pg. 6, para. 4; pg. 7, para. 4 and 6).
Further regarding claim 12¸ EL-Manzalawy does not disclose the following limitations:
EL-Manzalawy does not disclose the training of the MIL is at an instance-level, or that the predicting comprises predicting the label of new instances by applying the MIL classifier direction, or that the label of new bags comprising applying the MIL classifier to each instance of a respective bag and aggregation the results among all instances of the respective bag.
However, Park discloses a method of performing multiple instance regression for modeling peptides represented by MHC proteins (e.g. neoantigens) (Abstract; pg. 3033, para. 4), and discloses an instance-based type of multiple instance learning (pg. 3033, para. 2-3). Park discloses training a regression model at an instance level to make instant level predictions (pg. 3037, para. 2-5), and further discloses an aggregation function which combines the instance level predictions (pg. 3035, para. 6; pg. 3037, para. 4). Park discloses that meta-instance multiple instance learning methods (as used in EL-Manzalawy), unlike instance-level learning methods, cannot model individual instances or capture their relationship with bag-level responses, which is of great interest in many regression settings (pg. 3033, para. 2-3).
It would have been prima facie obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to have modified the multiple instance MIL of EL-Manzalawy to have used the instance-based multiple instance learning classifier to make instance level predictions and aggregate results of instances of a bag, as shown by Park above. One of ordinary skill in the art would have been motivated to combine the methods of EL-Manzalawy and Park in order to capture the relationship between individual instances and bag level responses, as shown by Park (pg. 3033, para. 2-3). This modification would have had a reasonable expectation of success given both EL-Manzalawy and Park use multiple instance learning to predict peptides interacting with MHC molecules, Park uses attributes of peptides in the instance-level modeling, and therefore the method of Park is applicable to EL-Manzalawy.
Therefore, the invention is prima facie obvious.
Claim 13 is rejected under 35 U.S.C. 103 as being unpatentable over El-Manzalawy in view of Park, as applied to claim 12 above, and further in view of Reynisson (2020).
Cited reference: Reynisson et al., Improved prediction of MHC II antigen presentation through integration and motif deconvolution of mass spectrometry MHC eluted ligand data, Feb. 2020, bioRxiv, pg. 1-34.
Regarding claim 13, El-Manzalawy discloses the method of claim 12 as applied above.
El-Manzalawy further discloses the training data includes a set of MHC-II molecules and experimentally determined binding affinities of MHC-II peptides to the MHC-II molecules (pg. 6, para. 2-4; Tables 1-3).
El-Manzalawy further discloses the MIL is trained to predict whether a particular peptide sequence will bind or not bind (i.e. be presented by ) the MHC molecules (pg. 6, para. 4; pg. 7, para. 4 and 6).
Further regarding claim 13, EL-Manzalawy in view of Park, as applied to claim 12 above, does not disclose that it is not known to which specific of the MHC molecules a peptide sequence is bound. Instead, EL-Manzalawy discloses the training data is experimentally determined affinities of MHC-II peptides to the MHC-II molecules (i.e. it is known which MHC-II molecule the binding affinity is for) (pg. 6, para. 2-4; Tables 1-3).
However, Reynisson discloses a method of improved prediction of MHC II antigen presentation through mass spectrometry eluted ligand data (Abstract), which includes training a machine learning model on large-scale eluted MHC II ligand mass spectrometry data to predict whether an antigen is presented by MHC II (i.e. a peptide binding to MHC II) (Abstract; pg. 6, para. 3 to pg. 7, para. 2; pg. 8, para. 3; pg. 9, para. 4, e.g. a pan-specific predictor of MHC II antigen presentation trained on EL data). Reynisson discloses the result of the mass spectrometry assay used in the training is a list of peptide sequences restricted to at least one of the MHC II molecules expressed by the interrogated cell line (i.e. it is not known which MHC molecule a peptide sequence binds in mass-spectrometry data) (pg. 3, para. 2). Reynisson discloses that most prediction methods are trained on either binding affinity (BA) data or MHC MS eluted ligand (EL) data (pg. 4, para. 2), and further discloses that evidence suggests that peptide MHCII binding affinity is a relative weak correlate of MHC antigen presentation, and several studies have demonstrated that MHC-II peptide prediction models can benefit from being trained on immunopeptidome data obtained by liquid chromatography coupled mass spectrometry (pg. 3, para. 2).
It would have been prima facie obvious to one of ordinary skill in the art, before the effective filing date of the claimed invention, to have modified the training data of EL-Manzalawy in view of Park, as applied to claim 12 above, to have used eluted MHC II ligand mass spectrometry data, as shown by Reynisson above. One of ordinary skill in the art would have been motivated to combine the methods of EL-Manzalawy in view of Park with Reynisson in order to provide improve performance of MHC-II peptide prediction models, given evidence suggests that peptide MHCII binding affinity is a relative weak correlate of MHC antigen presentation, as shown by Reynisson (pg. 3, para. 2). This modification would have had a reasonable expectation of success given Reynisson discloses binding affinity data (as used in EL-Manzalawy) or MS data are commonly used to make the same predictions of a peptide binding to an MHC-II molecule, and thus the data of Reynisson is applicable to the multiple instance learning method of EL-Manzalawy.
Therefore, the invention is prima facie obvious.
Conclusion
No claims are allowed.
Claim 7 is free of the prior art.
Claim 7 further limits the loss function used in training the MIL classifier to include a probability calibration production configured to calculate a probability of a previous training epoch in each training epoch, and wherein the probability is for a maximum of ŷi as max(fθ(xij), wherein xij corresponds to the tuple (si, ai). The closest prior art of record include EL-Manzalawy, Park, and Reynisson, applied in the above 103 rejection. Each of EL-Manzalawy and Park disclose methods of multiple instance learning for MHC peptide binding classification. However, the cited references do not disclose the specific loss function as recited in claim 7. It is noted the claim 7 is indefinite for the reasons discussed above, and reconsideration may be required in light of any amendments to claim 7 changing the scope of the claim.
Inquiries
Any inquiry concerning this communication or earlier communications from the examiner should be directed to KAITLYN L MINCHELLA whose telephone number is (571)272-6485. The examiner can normally be reached 7:00 - 4:00 M-Th.
Examiner interviews are available via telephone, in-person, and video conferencing using a USPTO supplied web-based collaboration tool. To schedule an interview, applicant is encouraged to use the USPTO Automated Interview Request (AIR) at http://www.uspto.gov/interviewpractice.
If attempts to reach the examiner by telephone are unsuccessful, the examiner’s supervisor, Olivia Wise can be reached at (571) 272-2249. The fax phone number for the organization where this application or proceeding is assigned is 571-273-8300.
Information regarding the status of published or unpublished applications may be obtained from Patent Center. Unpublished application information in Patent Center is available to registered users. To file and manage patent submissions in Patent Center, visit: https://patentcenter.uspto.gov. Visit https://www.uspto.gov/patents/apply/patent-center for more information about Patent Center and https://www.uspto.gov/patents/docx for information about filing in DOCX format. For additional questions, contact the Electronic Business Center (EBC) at 866-217-9197 (toll-free). If you would like assistance from a USPTO Customer Service Representative, call 800-786-9199 (IN USA OR CANADA) or 571-272-1000.
/KAITLYN L MINCHELLA/Primary Examiner, Art Unit 1685
Prosecution Insights