DETAILED ACTION
Applicant’s response, filed 19 Dec. 2025 has been fully considered. The following rejections and/or objections are either reiterated or newly applied. They constitute the complete set presently being applied to the instant application.
Notice of Pre-AIA or AIA Status
The present application, filed on or after March 16, 2013, is being examined under the first inventor to file provisions of the AIA .
Continued Examination Under 37 CFR 1.114
A request for continued examination under 37 CFR 1.114, including the fee set forth in 37 CFR 1.17(e), was filed in this application after final rejection. Since this application is eligible for continued examination under 37 CFR 1.114, and the fee set forth in 37 CFR 1.17(e) has been timely paid, the finality of the previous Office action has been withdrawn pursuant to 37 CFR 1.114. Applicant's submission filed on 19 Dec. 2025 has been entered.
Status of Claims
Claims 5, 17, 20-21, 26-27, 29-30, 32, and 35 are cancelled.
Claims 40-45 are newly added.
Claims 1-4, 6-16, 18-19, 22-25, 28, 31, 33-34, and 36-45 are pending.
Claims 8 and 22-23 are withdrawn from further consideration pursuant to 37 CFR 1.142(b) as being drawn to a nonelected species, there being no allowable generic or linking claim. Election was made without traverse in the reply filed on 05 July 2022.
Claims 1-4, 6-7, 9-16, 18-19, 24-25, 28, 31, 33-34, and 36-45 are rejected.
Priority
The effective filing date of the claimed invention is 07 Jan. 2021.
Claim Interpretation
Claim 1 recites “…panel-targeted sequencing”. Applicant’s specification at para. [00303], [00309], and [00319] discloses the panel-targeted sequencing can target one or more genomic regions, one or more chromosomes, or the whole-exome. Accordingly, “panel-targeted sequencing” is interpreted to be any sequencing method that targets less than the whole genome.
Claim 1 recites “E) applying…(i) all or a portion of the first mapped dataset and (ii) all or a portion of the second mapped dataset to a model, wherein the model includes a segmentation step”. In light of Applicant’s specification at FIG. 4B and [00251], which discloses that applying a model can include applying two separate component models or a joint model, the applying component models of a larger model is considered to read on the claims.
Claim 31 recites “…wherein the first biological sample and the second biological sample are the same biological sample”. In light of Applicant’s specification at para. [00293], the claim is interpreted to mean that the first and second biological samples are two samples from the same biological sample.
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.
Claim 45 is rejected under 35 U.S.C. 112(b) as being indefinite for failing to particularly point out and distinctly claim the subject matter which the inventor or a joint inventor regards as the invention. This rejection is newly recited and necessitated by claim amendment.
Claim 45 is indefinite for recitation of “the breakpoint is between the first bin and the second bin in the reference genome”. Claim 1, from which claim 45 depends recites “a second bin in the first plurality of bins mapping to the first genomic region” and “a first genomic region that maps to the breakpoint”. Given the first genomic region, to which the second bin maps, maps to the breakpoint, it is unclear in what way the breakpoint is intended to be between the first bin and the second bin. In other words, it is not clear if the breakpoint is within the first genomic region, as suggested by claim 1, or if the breakpoint is between the first bin and the first genomic region (e.g. the second bin). Clarification is requested.
Claim Rejections - 35 USC § 103
The rejection of claim 35 under 35 U.S.C. 103 in the Office action mailed 23 July 2025 has been withdrawn in view of the cancellation of this claim received 19 Dec. 2025.
In the event the determination of the status of the application as subject to AIA 35 U.S.C. 102 and 103 (or as subject to pre-AIA 35 U.S.C. 102 and 103) is incorrect, any correction of the statutory basis (i.e., changing from AIA to pre-AIA ) for the rejection will not be considered a new ground of rejection if the prior art relied upon, and the rationale supporting the rejection, would be the same under either status.
The following is a quotation of 35 U.S.C. 103 which forms the basis for all obviousness rejections set forth in this Office action:
A patent for a claimed invention may not be obtained, notwithstanding that the claimed invention is not identically disclosed as set forth in section 102, if the differences between the claimed invention and the prior art are such that the claimed invention as a whole would have been obvious before the effective filing date of the claimed invention to a person having ordinary skill in the art to which the claimed invention pertains. Patentability shall not be negated by the manner in which the invention was made.
This application currently names joint inventors. In considering patentability of the claims the examiner presumes that the subject matter of the various claims was commonly owned as of the effective filing date of the claimed invention(s) absent any evidence to the contrary. Applicant is advised of the obligation under 37 CFR 1.56 to point out the inventor and effective filing dates of each claim that was not commonly owned as of the effective filing date of the later invention in order for the examiner to consider the applicability of 35 U.S.C. 102(b)(2)(C) for any potential 35 U.S.C. 102(a)(2) prior art against the later invention.
Claims 1-4, 6-7, 9-16, 18, 24-25, 31, 33-34, 36, 38, and 44-45 are rejected under 35 U.S.C. 103 as being unpatentable over Harris (2015), as evidenced by Ghahramani (2001) in view of Talevich (2016). Any newly recited portion is necessitated by claim amendment.
Cited references:
Harris et al., US 2015/0066824 A1, published 5 March 2015, and effectively filed 29 Aug. 2014 (previously cited);
Ghahramani, An Introduction to Hidden Markov Models and Bayesian Network, 2001, International Journal of Pattern Recognition and Artificial Intelligence, 15(1), pg. 9-42 (previously cited); and
Talevich et al., CNVkit: Genome-Wide Copy Number Detection and Visualization from Targeted DNA Sequencing, PLOS Computational Biology, 2016, pg. 1-18 (previously cited);
Regarding claim 1, Harris discloses the method for determining a copy number variation status of a human subject (Abstract; [0009]; [0197]) comprises the following steps:
Harris discloses A) performing whole-genome sequencing on a first plurality of DNA molecules from a first biological sample of the subject to produce a first plurality of untargeted nucleic acid sequences ([0005], receive untargeted sequencing data from first nucleic acid sample; [0009]; [0056], WGS covers 99.99% of genome; [0085])); [0031], nucleic acids include DNA), at an average sequencing depth of 1X (i.e. from 0.5 to 5X) across 99% of a reference genome ([0009; [0056], WGS covers 99.99% of genome; [0085]), and receiving the first plurality of nucleic acid sequences electronically ([0005], e.g. sequencing data received by computer). Harris et al. further shows the first plurality of sequences comprises at least 100,000 nucleic acid reads (i.e. sequences) ([0098]; [0100]).
Harris discloses B) performing panel-targeted sequencing, including whole-exome sequencing on a second plurality of DNA molecules from a second biological sample of the subject to obtain a second plurality of nucleic acid sequences ([0005]; [0011]; [0031]; [0048]) and receiving the second plurality of nucleic acid sequences electronically ([0005];[0147]), wherein the second plurality of sequencing data comprises at least 10,000 nucleic acid sequences ([0068]);
Harris discloses C) aligning (i.e. mapping), using the computer ([0147]), the first plurality of nucleic acid sequences to positions within a first reference sequence from a human (i.e. mapping to positions within a reference sequencing of the same species as the subject) ([0007]). Harris further discloses copy number variations are detected over the entire genome and the coverage of the first sequencing data can be 99.9999% of the whole genome (i.e. the sequencing data maps to 99.9% of the whole genome) ([0085]; [0340]), which shows the first reference sequence can be a reference genome. Given the reads are whole genome sequencing reads aligned to the genome, this demonstrates this includes a first genomic region covering the breakpoint. Furthermore, Harris discloses the panel-targeted sequencing is whole-exome sequencing ([0011]; [0048]), which necessarily would not include any introns (and thus the breakpoint in the intron). Harris discloses the whole genome sequencing data is assigned into genomic bins with measured amounts of sequence reads in each bin (i.e. an identify of a first plurality of bins and a bin value for each bin in the first plurality of bins) ([0009]; [0059]).
Harris discloses D) aligning (i.e. mapping), using the computer ([0147]), the second plurality of nucleic acid sequences from the whole-exome sequencing to positions within a second reference sequence (i.e. a reference construct), including a reference genome ([0010], first and second reference sequences can be the same; [0340], reference sequence can be a reference genome), which includes the plurality of genomic regions targeted by the panel-targeted sequencing, thereby obtaining mapped on-target nucleic acid sequences in the second mapped dataset, including the first exon given the second plurality of sequences were from whole exome sequencing.
Harris discloses E) applying, using the computer ([0147]), the first mapped data and the second mapped data set to one or more statistical models, including a Hidden Markov Model (HMM) (Figure 2B-C, e.g. assay 1 and 2 applied to same or separate models C; [0010]; [0016] ;[0079]; [0340]), wherein each model identifies a variant including a copy number variation indicating the copy number variation status, including a deletion, in regions of the subject ([0075]-[0076], e.g. model output is copy number variations; [0079], e.g. multiple algorithms/models used depending on data inputs to identify variants; [0082]; claim 7, e.g. deletions detected). Harris further discloses the statistical model detects CNVs “segmentally” ([0083]) and involves assigning data into genomic bins (i.e. segments) ([0009]). Harris discloses the genomic regions analyzed can include an exon (i.e. a deletion in an exon is determined) and intron ([0222]-[0223]) and also genome-wide structural variations with a sensitivity of less than 10 kb ([0060]). Harris further discloses the model segments the sequencing data into different regions by assigning the combined whole genome and targeted sequencing data (i.e. the first and second data) with redundant sequences removed into genomic bins ([0007]; [0010] and [0014], e.g. combined data set assigned to bins and used to generate output). Harris further discloses the model analyzes the number of sequence reads per bin ([0332]; [0337]; [0340]), including reads mapped to detection breakpoints ([0010], e.g. HMM measures reads near detected feature’s breakpoint; [0085]), in order to detect one or more genomic regions including copy number variations ([0079]; [0082]; [0187; claim 7).
Regarding claim 1, Harris does not disclose the following:
Regarding claim 1, Harris does not disclose the panel targeted sequencing uses an average sequencing depth of between 50X-500X across the genomic regions targeted by the panel-targeted sequencing.
However, Talevich discloses a method for genome-wide copy number detection using targeted reads and nonspecifically captured off-target reads within targeted DNA sequencing data (Abstract), and discloses that performing exome (i.e. targeted-panel) sequencing at a higher coverage increases the sensitivity for calling variants in clinical use (pg. 2, para. 1), and discloses that performing the read depth calculation is the most computationally demanding step, and for exome sequencing data at 100-fold coverage (i.e. between 50X and 500X across the targeted genomic regions) takes on the order of 20 minutes (pg. 3, para. 2).
Therefore, 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 utilized 100X coverage across the regions targeted by the panel-targeted sequencing through routine experimentation of the sequencing coverage within the prior art conditions of increasing coverage to increase the sensitivity for calling variants for clinical use and decreasing coverage to reduce computational run time. See MPEP 2144.05 II. A.
Further regarding claim 1, Harris does not disclose the second mapped dataset comprises an identity of a second plurality of bins across the reference construct, and a bin value or a copy number state for each bin in the second plurality of bins (instead Harris discloses bin values determined from read coverages of a combined data set from the first and second mapped read sets ([0010]). Furthermore, Harris, does not explicitly disclose the segmentation step assigns, to a single segment, the bin value or the copy number state of a first bin in the second plurality of bins mapping to the first exon and the bin value or the copy number state of a second bin in the first plurality of bins mapping to the first genomic region, thereby identifying the copy number variation status of the single exon copy number variation for the first exon of the gene as output. Similarly, Harris does not explicitly disclose detecting a copy number variation status of a single exon associated with a breakpoint in an intron, or that the intron includes the breakpoint.
Instead, Harris discloses applying the model to a combined dataset containing bin values of mapped reads of the first and second datasets to a statistical model that detects CNVs in exons and/or introns “segmentally” and involves assigning the combined whole genome and targeted sequencing data (i.e. the first and second data) with redundant sequence information removed into genomic bins (Figure 2B-C, e.g. assay 1 and 2 applied to same or separate models C; [0007], [0010] and [0014], e.g. e.g. combined data set assigned to bins and used to generate output). [0010]; [0016] ;[0079]; [0222]-[0223], e.g. introns and exons; [0340]), without explicitly specifying the two specific bins/segments corresponding to an intron region and exon region and the specific assigning of the first bin and second bin as claimed.
However, these limitations are obvious in view of Talevich.
As discussed above, Talevich discloses a method for genome-wide copy number detection using targeted reads (e.g. the second mapped dataset) and nonspecifically captured off-target reads, which provide a very low-coverage sequencing of the whole-genome (pg. 2, para. 2) (e.g. the first mapped dataset), within targeted DNA sequencing data (Abstract). Talevich discloses the method comprises determining coverages in both off-target bins and on-target bin to determine bin coverages and then bin-level log2 ratios (i.e. a dataset of bin values or copy number states for each of the first and second reads are determined), and estimating a copy numbers using the target and off-target bin specific log2 ratios (i.e. segmentation using the first and second copy number states/bin values) (Figure 1, pg. 3, para. 1; pg. 7, para. 4). Talevich specifically discloses deviating from a user-specified bin size to use bins that fit into small regions such as introns (pg. 4, para. 2), such that copy number changes in introns can also be detected. Talevich discloses this combination achieves both exon-level resolution in targeted gene regions and sufficient resolution in the larger intronic regions to identify copy number changes, and successfully inferred copy numbers at 100 kb resolution genome-wide (Abstract; pg. 2, para. 1; Fig. 5). Accordingly the off-target bins include counts of reads mapped to introns, including any breakpoints, and the target bins include reads mapped to exons of the targeted genes or exome. Talevich further discloses determining a whole-genome copy number profile including insertions and deletions within genes with the identified breakpoints, which are inherently either in an intronic or exonic region of the respective gene (Fig. 5A and B). Talevich further discloses that the off-target regions are between each target bin (pg. 4, col. 1), demonstrating the on-target and off-target bins are not overlapping. Therefore, any detected deletion (i.e. single segment) with a breakpoint in an intron (i.e. in an off-target region) detected by the model includes the assigned copy number value of the on-target bin (i.e. in the exon) and the assigned copy number value of the off-target bin (i.e. in the intron). Talevich discloses the method provides highly accurate and reliable copy number estimates genome-wide (pg. 2, para. 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 method of Harris to have determined a count of off-target reads in a region containing an intron, a count of off-target regions containing an exon, and detected a copy number variant of an exon with a breakpoint in an intron in a genome-wide copy number profile by assigning the corresponding bin value/copy numbers state of off-target and on-target bins to the same segment (i.e. the deletion), as shown by Talevich, given Talevich discloses identifying all deletions/insertions and breakpoints within genes and using bins that fit into introns (Abstract; Figure 1 and 5; pg. 2, para. 2; pg. 3, para. 1; pg. 4; pg. 7, para. 4) One of ordinary skill would have been motivated to further combine the methods of Harris and Talevich to detect all copy number variants in all genes, including deletions/insertions in exons with breakpoints in introns or exons in a genome-wide copy number profile, in order to provide genome-wide copy number profiles and maximize the copy number information obtained from targeted sequencing, as shown by Talevich (pg. 2, para. 3). This modification would have had a reasonable expectation of success given both Harris ([0060]) and Talevich use targeted sequencing reads and lower coverage off-target reads (i.e. whole genome sequencing data) to determine copy numbers genome-wide, such that the model of Talevich is applicable to Harris, and furthermore, given Talevich discloses being able to detect breakpoints and copy number variations within genes. Furthermore, Talevich discloses the off-target reads provide low-coverage sequencing of the whole genome (pg. 2, para. 2), such that the low-coverage whole genome sequencing data of Harris can be used in the method of Talevich in place of the off-target reads in off-target bins.
Regarding claim 2, Harris further discloses the whole-genome sequencing may comprise at least 10,000,000 sequencing reads ([0098];[0100]).
Regarding claim 3¸ Harris further discloses the whole-genome sequencing data (i.e. the first plurality of at least 100,000 nucleic acid sequences) can provide a sequencing depth of 2X or 3X across the genome of the subject ([0061], WGS data can comprise second or third reads; [0332] and FIG. 4, e.g. “single read” WGS refers to 1X sequencing depth).
Regarding claim 4, Harris further discloses the second plurality of at least 10,000 nucleic acid sequences includes at least 100,000 sequence reads ([0068]).
Regarding claim 6, Harris further discloses the panel-targeted sequencing is whole exome sequencing, which includes at least 25 genes ([0011]; [0048]; [0050]).
Regarding claim 7, Harris further discloses the panel-targeted sequencing is whole exome sequencing ([0011]; [0048]).
Regarding claim 9, Harris further discloses the first and second biological samples may be a blood or saliva sample ([0199]), and that the first and second samples can be different samples (i.e. independently selected) ([0006]).
Regarding claim 10¸ Harris further discloses using the measured counts per bin (i.e. the first respective bin values) of the whole-genome sequencing data in the model, wherein the model is a hidden markov model, to detect copy number ([0010], e.g. model is hidden Markov model; [0332]-[0334]; [0340], e.g. hidden markov model uses read bin counts in whole genome sequencing data to predict copy number).
Regarding claim 12, Harris further discloses, for the targeted sequencing data (i.e. the second mapped dataset), measuring the number of sequence reads (i.e. a respective second bin value) mapped to each of a plurality of genomic bins representing unique segments of the reference construct ([0010]; e.g. data from the combined dataset is set into genomic bins; [0059], e.g. targeted-specific sequencing data assigned into bins; FIG. 5, e.g. bins are non-overlapping)
Harris further discloses using the measured counts per bin (i.e. the second respective bin values) of the targeted sequencing data in the model to detect copy number ([0010], e.g. model used to generate copy number output; FIG. 2)
Regarding claims 11 and 13:
First, regarding claim 13, Harris further discloses, for the targeted sequencing data (i.e. the second mapped dataset), measuring the number of sequence reads mapped to each of a plurality of genomic bins representing unique segments of the reference construct ([0010]; 0059], e.g. targeted-specific sequencing data assigned into bins; FIG. 5, e.g. bins are non-overlapping).
Further regarding claims 11 and 13, Harris does not disclose the following:
Regarding claim 11, Harris, as applied to claim 1 above, does not disclose determining a respective copy number state for each respective bin in the first plurality of bins using the respective first bin value for the respective bin; and the all or the portion of the first mapped dataset inputted into the model in E) comprises the respective copy number state for each respective bin in the first plurality of bins.
Further regarding claim 13¸ Harris, as applied to claim 1 above, does not disclose determining a respective copy number state for each respective bin in the second plurality of bins using the respective second bin value for the respective bin; and the all or the portion of the second mapped dataset inputted into the model in E) comprises the respective copy number state for each respective bin in the second plurality of bins.
However, regarding claims 11, and 13, as discussed above, Talevich discloses a method for genome-wide copy number detection using targeted reads and nonspecifically captured off-target reads within targeted DNA sequencing data (Abstract), which comprises determining coverages in both off-target bins and on-target bins (i.e. determining first and second respective values for a first and second plurality of bins) (figure 1, pg. 3, para. 1). Talevich further discloses the methods includes estimating a copy number ratio using the target and off-target bins (i.e. copy number states for each first and second bin using the respective first and second bin values) (Figure 1, pg. 3, para. 1), and then inputting the copy number ratios for the first and second plurality of bins, which are non-overlapping as discussed above for claim 1 (i.e. the copy number states) into a circular binary segmentation (CBS) model to predict one or more copy number variations in a subject (Figure 1; pg. 7, para. 4-5). Talevich further discloses the off-target reads provide a very low-coverage sequencing of the whole-genome (pg. 2, para. 2). Last, Talevich discloses the model provides highly accurate and reliable copy number estimates genome-wide (pg. 2, para. 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 method of Harris in view of Talevich to have determined a copy number state using the first and second respective bin values (e.g. the off-target and on-target bin coverages, respectively) and then inputted the copy number states of the first and second plurality of bins into a respective model to determine one or more copy number variations, as shown by Talevich (Figure 1 and 5; pg. 3, para. 1; pg. 4, para. 2; pg. 7, para. 4-5), thus arriving at the invention of claims 11, and 13. One of ordinary skill in the art would have been motivated to combine the methods of Harris and Talevich in order to provide reliable copy number estimates, as shown by Talevich (pg. 2, para. 3). This modification would have had a reasonable expectation of success because Harris also discloses using model with bin coverages to estimate copy number, such that the model of Talevich is applicable to Harris.
Further regarding claim 14, Harris further discloses the number of bins can include 1000, 1kb bins (i.e. 1000 bins, each less than 5kb) ([0059], can include 1kb bins; FIG. 5, e.g. 1kb bins provide 1000 bins).
Further regarding claim 15, Harris further discloses the bins can include 1000, 1 kb bins ([0059]; FIG. 5), such that the second plurality of bins collectively represents more than 10kb of the reference construct.
Further regarding claim 16, Harris further discloses the bins can be 1 kb bins (i.e. each respective bin corresponds to no more than 1kb of the reference construct ([0059]; FIG. 5).
Regarding claim 18, Harris further discloses the first mapped dataset represents a number of sequence reads mapped to each of a plurality of 10k bins ([0059]), which shows the first mapped dataset collectively represents sequencing depths (i.e. the number of mapped reads) for at least 10kb of the reference genome, given each bin must be at least 1 base pair long.
Harris further discloses the second mapped dataset represents a number of sequence reads mapped to each of a plurality of 10k bins ([0059]), which shows the second mapped dataset collectively represents sequencing depths (i.e. the number of mapped reads) for at least 10kb of the reference construct.
Regarding claim 24, Harris further discloses the model determines copy number variation status of the entire genome through a hidden markov model (i.e. through statistical inference) ([0010]; [0337]; [00340]).
Regarding claim 25, Harris further discloses the model comprises a hidden markov model (([0010]; [0337]; [00340]), which is a type of probabilistic network, as evidenced by Ghahramani. Ghahramani discloses that hidden markov models are a particular kind of Bayesian (i.e. probabilistic) network (pg. 1, para. 1, second bullet). Accordingly, Harris discloses the model comprises a probabilistic network.
Regarding claim 31¸ Harris further discloses the first biological sample and the second biological sample are the same sample ([0006]; [0010]).
Regarding claims 33-34¸ Harris discloses the panel-targeted sequencing targets specific genes (i.e. a gene-panel) or is whole-exome sequencing ([0009]; [0011])
Further regarding claims 33-34¸Harris does not disclose the panel-targeted sequencing is at an average sequencing depth of between 60X and 80X across the genome or an average depth of between 200X and 300X across the targeted genomic regions.
However, Talevich discloses a method for genome-wide copy number detection using targeted reads and nonspecifically captured off-target reads within targeted DNA sequencing data (Abstract), and discloses that performing exome or a gene target panel sequencing at a higher coverage increases the sensitivity for calling variants in clinical use (pg. 2, para. 1), and discloses that performing the read depth calculation is the most computationally demanding step, and for exome sequencing data at 100-fold coverage or a 293-gene target panel at 500-fold coverage takes on the order of 20 minutes (pg. 3, para. 2).
Therefore, 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 utilized lower coverages, such as 60X or 80X or higher coverages at 200X and 300X coverage across the regions targeted by the panel-targeted sequencing through routine experimentation of the sequencing coverage within the prior art conditions of increasing coverage to increase the sensitivity for calling variants for clinical use and decreasing coverage to reduce computational run time. See MPEP 2144.05 II. A.
Regarding claim 36, Harris discloses variants detected in the low coverage whole genome sequencing data (i.e. first mapped dataset) can be used to impute variants over the whole genome ([0055]; [0337]).
Harris further discloses determining a genetic risk of a disease based on the sets of outputs ([0176]-[0177[), which include the imputed variants (([0055]; [0337]).
Regarding claim 38, Harris in view of Talvich make obvious the exon is of particular genes. Talevich discloses these genes are cancer-relevant genes and include “MET” (Figure 5; pg. 11, para. 4), associated with breast and/or ovarian cancer, such that any CNV detected in the gene provides in indication of cancer in the subject.
Regarding claim 44¸ Harris in view of Talevich disclose applying (i) all or a portion of the first mapped dataset and (ii) all or a portion of the second mapped dataset to a model, as applied to claim 1 above. Given the bin coverages and thus copy ratios of the target bins and off-target bins of Talevich (Fig. 1) are non-overlapping (pg. 4, para. 1), whether the data is applied to the model separately or combined in one data set 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 inputting the non-overlapping data together in one dataset, as shown by Talevich. Therefore, the data of Talevich would perform equally as well in determining copy number variations across the whole genome, including in exons, and such a modification fails to patentably distinguish over Harris in view of Talevich.
Regarding claim 45, Harris in view of Talevich make obvious using bin and copy number values of target bins and off-target bins to achieve exon-level resolution in targeted gene regions and sufficient resolution in large intronic regions to identify copy number changes genome-wide (Tavelvich; Abstract; pg. 2, para. 1; Fig. 5). Given Talevich performs genome-wide copy number calling with exon level resolution, this shows the model maps a third on-target bin (i.e. a third bin in the second plurality of bins) to a second exon within a gene. Talevich further discloses the model assigns bins to segments according to similar log2ratios (i.e. bins assigned to the single segment have similar coverages) (pg. 7, para. 4-5; Fig. 5), and discloses determining a whole-genome copy number profile including insertions and deletions within genes with the identified breakpoints, which are inherently either in an intronic or exonic region of the respective gene with the intron between the two exons (i.e. the breakpoint between the first bin and the second bin) (Fig. 5A and B).
Therefore, the invention is prima facie obvious.
Claims 19 and 28 are rejected under 35 U.S.C. 103 as being unpatentable over Harris (2015) in view of Talevich, as applied to claim 1 above, further in view of Lachlan (2012), as evidenced by Haugh (2017). This rejection is previously recited.
Cited references:
Lachlan et al, An exome sequencing pipeline for identifying and genotyping common CNVs associated with disease with application to psoriasis, 2012, ECCB, 28, pg. i370-i374 (previously cited); and
Haugh 2017, Machine Learning for OR & FE: Hidden Markov Models, 2017, Columbia University, pg. 1-34; Pub. Date: 2017, Wayback Machine (previously cited).
Regarding claims 19 and 28¸ Harris in view of Talevich disclose the method of claim 1 as applied above.
Further regarding claims 19 and 28¸ Harris in view of Talevich does not explicitly disclose the model comprises at least 500 parameters, as recited in claim 19, or that the model processes the (i) all or the portion of the first mapped dataset and (ii) all or the portion of the second mapped dataset to identify the one or more copy number variations as output of the model in N-dimensional space in the applying E), wherein N is a positive integer of 4 or greater. However, these limitations were obvious to one of ordinary skill in the art for the following reasons.
Harris discloses applying a hidden markov model to a combined dataset comprising the first and second mapped datasets to identify one or more copy number variations indicating the copy number variation status of the subject ([0010], wherein the output is generated by a statistical model; [014]; [0071]; [0337]; claims 1 and 7). Harris further discloses the first and second mapped datasets can include read counts within genomic bins, and the number of genomic bins can include 1000 bins ([0059]; [0337]; [0340]). Harris further discloses that methods that reduce a number of variables such as a principal component analysis (PCA) can be used in the analysis of the data ([0082]), suggesting the application of PCA to the first and second mapped datasets.
Furthermore, Lachlan discloses a method for identifying copy number variations in a DNA sample of a subject (Abstract), which comprises performing principal component analysis on read depth information for each region of N genomic regions (i.e. each bin of a plurality of bins) to calculate the first 50 components (i.e. dimensionality reduction components) (pg. i371, col. 1, para. 7), projecting the first 40 principal components, and modeling the absolute copy numbers of the sample by applying a hidden markov model to the first 40 principal components (i.e. the model is applied in 40-dimensional space on the read depth information, as recited in claim 28), as recited in claim 28 (pg. i371, col. 2, para. 3-5; Figure 1; Table 3). Lachlan further discloses the hidden markov model uses a different emission distribution per hidden copy number genotype per 100-bp window to model the normalized read depths determined by PCA (pg. i371, col. 2, para. 5-6). Given Harris discloses the number of genomic bins can include 1000 bins ([0059]), the hidden markov applied to read depth data of 1000 bins would necessarily comprise at least 1000 parameters (i.e. at least 500 parameters, as recited in claim 19), as evidenced by Haugh. Haugh discloses a hidden markov model is defined by its emission distribution (pg. 5), and that a parameter Bij for the emission distribution is estimated. Accordingly, the hidden markov model generating an emission distribution per copy number state and per 100 bp window, for 1000 windows, would result in a hidden markov model comprising at least 1000 parameters (e.g. at least one per window). Lachlan further discloses that principal components correct for read-depth biases from sequencing data (pg. i373, col. 1, para. 2), and that the principal component algorithm has a very low memory footprint and is highly computationally efficient (pg. i371, col. 1, para. 7).
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 made obvious by Harris in view of Talevich, as applied to claim 1 above, to have applied a hidden markov model to the top 40 dimensionality reduction components of the read depth information by generating an emission distribution per copy number state and genomic bin, to estimate a copy number variation status of the subject, according to the method of Lachlan (pg. i371, col. 1, para. 7 and col. 2, para. 3-6; Figure 1), resulting in an application of a hidden markov model to with greater than 500 parameters to read depth information in greater than 4-dimensional space, as discussed above, thus arriving at the inventions of claims 19 and 28. One of ordinary skill in the art would have been motivated to combine the methods of Harris in view of Talevich and Lachlan in order to correct for read-depth biases in the sequencing data using a highly computationally efficient algorithm before predicting copy number variations, as disclosed by Lachlan (pg. i371, col. 1, para. 7; pg. i373, col. 1, para. 2). This modification would have had a reasonable expectation of success because both Lachlan and Harris analyze sequencing depth information in genomic bins, and Harris shows using a hidden markov model to predict copy number (FIG. 4; [0340]) and that a principal component analysis can be used to analyze the sequencing data ([0082]) , such that the method of Lachlan is applicable to the sequencing data of Harris.
Therefore, the invention is prima facie obvious.
Claim 37 is rejected under 35 U.S.C. 103 as being unpatentable over Harris in view of Talevich, as applied to claim 36 above, and further in view of Chen (2020). This rejection is previously recited.
Cited reference: Chen et al., Genotype imputation and variability in polygenic risk score estimation, 2020, Genome Medicine, 12:100, pg. 1-13; previously recited.
Regarding claim 37¸ Harris in view of Talevich disclose the method of claim 36 as applied above.
Further regarding claim 37, Harris in view of Talevich, as applied to claim 36 above, do not disclose the disease risk is a polygenic risk score.
However, Chen discloses a method for estimating polygenic risk scores using imputed genotypes (Abstract), which comprises imputing variants (pg. 3, col. 2, para. 2) and then calculating a polygenic risk score using the imputed variants (pg. 4, col. 1, para. 1 to col. 2, para. 1). Chen further discloses polygenic risk scores can be used to guide healthcare decisions (Abstract).
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 Harris in view of Talevich to have calculated a polygenic risk score based on imputed variants, as shown by Chen, discussed above. One of ordinary skill in the art would have been motivated to combine the methods of Harris in view of Talevich and Chen in order to guide healthcare decisions of the subject, as shown by Chen (Abstract). This modification would have had a reasonable expectation of success given Harris discloses determining imputed variants, such that the calculations of Chen are applicable to the data of Harris.
Therefore, the invention is prima facie obvious.
Claim 39 is rejected under 35 U.S.C. 103 as being unpatentable over Harris in view of Talevich, as applied to claim 1 above, and further in view of Azrak (2015). This rejection is previously recited.
Cited reference: Azrak, upQMPSF, a Method for the Detection of BRCA Exon Copy Number Variants, 2015, Biochem Genet, 53:141-157; previously recited.
Regarding claim 39, Harris in view of Talevich disclose the method of claim 1 as applied above.
Further regarding claim 39, Harris in view of Talevich do not disclose the first exon is in BRCA1 or BRCA 2.
However, Azrak discloses a method for detecting a BRCA1 exon copy number variant and discusses that large insertion/deletion mutations are frequently found in genes associated with certain diseases such as cancer (Abstract). Azrak further discloses a high frequency of large genomic rearrangements is reported in BRCA1 in hereditary breast and ovarian cancer patients, including rearrangements spanning one or more exons (pg. 141, 1, para. 1 to pg. 142, para. 1). Azrak discloses these germline mutations in BRCA1 and BRCA2 genes account for most familial breast and ovarian cancer cases (pg. 155, para. 4) and that methods for detecting large copy number variations can be used as a mutation screening assay in genetic testing for hereditary diseases (pg. 155, para. 5-7).
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 Harris in view of Talevich, as applied to claim 1 above, to have detected the copy number variation in a BRCA1 or BRCA2 exon, as shown by Azrak, discussed above. One of ordinary skill in the art would have been motivated to combine the methods of Harris in view of Talevich with Azrak in order to screen for hereditary diseases, given BRCA1 and BRCA2 account for most familial breast and ovarian cancer cases, as shown by Azrak (pg. 155, para. 4-7). This modification would have had a reasonable expectation of success because Harris in view of Talevich disclose a method of generating a whole genome copy number profile, including detecting copy number variants within genes, and thus the method of Azrak is applicable to the copy number detection method of Harris in view of Talevich.
Therefore, the invention is prima facie obvious.
Claims 40-43 are rejected under 35 U.S.C. 103 as being unpatentable over Harris in view of Talevich, as applied to claim 1 above, and further in view of Garvin (2015).. This rejection is newly recited and necessitated by claim amendment.
Cited reference: Garvin et al., Interactive analysis and assessment of single-cell copy-number variations, 2015, Nature Methods, 12(11), pg. 1058-1060 and ONLINE Methods; newly cited.
Regarding claim 40, Harris in view of Talevich make obvious the method of claim 1 as applied above.
Further regarding claim 40, Harris in view of Talevich does not disclose the first plurality of bins and the second plurality of bins are each variable-sized bins in which bin size is selected to contain a constant number of mappable positions across the first and second plurality of bins, respectively.
However, Garvin discloses a method for analyzing copy-number variations using sequencing data (Abstract; pg. 1058, col. 1, para. 1), which includes binning aligned reads into variable-length intervals across the genome that contained the same number of uniquely mappable positions (pg. 1059, col. 2, para. 1 to pg. 1060, col. 1, para. 1; pg. 1 of ONLINE Methods: Binning Method). Garvin discloses that copy number analysis begins with binning across mapped reads into fixed or variable length intervals across the genome, which aggregates read-depth information into regions that are more robust, and that fixed length bins are discouraged because they lead to read dropout in regions that span repetitive and complex regions (pg. 1 of ONLINE METHODS, col. 1, para. 2-4, e.g. “Binning method”).
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 Harris in view of Talevich, as applied to claim 1 above, to have used the variable-sized bins of Garvin, discussed above. One of ordinary skill in the art would have been motivated to combine the methods of Harris in view of Talevich with Garvin in order to avoid using fixed length bins that lead to read dropout in regions that span repetitive sequences or complex regions, as shown by Garvin (pg. 1 of ONLINE METHODS, col. 1, para. 2-4, e.g. “Binning method”). This modification would have had a reasonable expectation of success given Harris in view of Talevich bin first and second read sets for copy number analysis, and thus the binning method of Garvin is appliable to Harris in view of Talevich.
Regarding claims 41-43, Harris in view of Talevich make obvious using the first and second mapped datasets containing bin values or copy number states in a model as claimed, as applied to claim 1 above.
Regarding claims 41-42, Talevich further discloses correcting for biases at each bin by correcting for repeat-masked fraction (i.e. predicted mappability) and GC content (pg. 7, para. 3).
Regarding claim 43¸ Talevich further discloses the coverage of each bin is divided by the size of the bin (pg. 4, para. 3).
Regarding claims 41-43, is further noted that the embodiment of the mapped datasets having “bin values” is recited as an alternative (i.e. not required), and Talevich discloses the copy number state (i.e. log2 ratios) of the bins as claimed (Fig. 1, pg. 7, para. 4).
Therefore, the invention is prima facie obvious.
Response to Arguments
Applicant's arguments filed 06 June 2025 regarding 35 U.S.C. 102/103 have been fully considered but they are not persuasive.
Applicant overviews the claims and prior art at pg. 13, para. 6 to pg. 16, para. 1, and further remarks that Harris combines the WGS and WES prior to segmentation, and thus in Harris, there is no joint segmentation using both the first and second mapped datasets with their mapping contexts preserved, and among the cited Harris paragraphs, only certain paragraphs make reference to using two different datasets to identify features, but these paragraphs do not support the proposition of using the two datasets with the mapping context of the first and second mapped datasets preserved (Applicant’s remarks at pg. 16, para. 2-3). Applicant remarks that the reads in Harris are pooled, and bins are created over the pooled read counts and the mapping contexts associated with the targeted and untargeted data no longer exist independently, while claim 1 in contrast, assigns one bin from each dataset to the same segment and the WGS bins and panel derived bins maintain their separate coverage distributions and mapping contexts and performs segmentation on unmerged datasets (Applicant’s remarks at pg. 16, para. 4 to pg. 17, para. 4). Applicant further remarks that various paragraphs of Harris do not remedy the deficiencies discussed above (Applicant’s remarks at pg. 17, para. 5 to pg. 19, para. 1).
This argument is not persuasive. In response to applicant's arguments against the references individually, one cannot show nonobviousness by attacking references individually where the rejections are based on combinations of references. See In re Keller, 642 F.2d 413, 208 USPQ 871 (CCPA 1981); In re Merck & Co., 800 F.2d 1091, 231 USPQ 375 (Fed. Cir. 1986).
It is agreed that Harris, including the paragraphs cited by Applicant, does not disclose having separate datasets of bin values of the first mapped read set and bin values of the second mapped read sets as input into the model. However, the model of Talevich being combined with Harris does make obvious this feature. Talevich discloses having separate coverage files for target bins and off-target bins (Fig. 1), with these off-target regions being between the on-target regions (i.e. non-overlapping) (pg. 4, para. 1). This corresponds to the different “mapping contexts” referred to by Applicant. The bin coverages of the off-target bins and the on-target bins are converted to copy number ratios (i.e. a copy number state for each bin) (Fig. 1), and thus each on-target bin and each off-target bin has a copy number ratio (i.e. the first and second mapped datasets).
It is further noted that while Applicant remarks segmentation is performed on unmerged datasets, this argument is not commensurate with the scope of the independent claim 1. Claim 1 only requires that the model is applied to the first and second mapped datasets, which each contain the bin information; however, nothing in the claims precludes the bins from having been merged into a single file. The claims only specify how the assigning of bin values or copy number states occurs; given Talevich performs segmentation on copy ratios of distinct bin sets (i.e. from , each bin is assigned to a given segment. As explained in the above rejection, Talevich discloses detecting copy number variations at single exon resolution across the genome, which demonstrates the assigning of a bin or copy number state of a first bin mapping to the first exon and a second bin mapping to the first genome region containing the breakpoint.
Applicant remarks that Harris’s “segmental” CNV detection and assigning data into genomic bins is incorrect because Harris describes a long CNV covered by a single sequencing modality resulting in a single coverage profile, and nothing in para. [0083] suggests the use of two sequencing modalities (Applicant’s remarks at pg. 19, para. 2-3). Applicant further overviews various paragraphs of Harris explaining why Harris does not disclose an algorithm handing multiple sequencing datasets or bins from multiple sequencing modalities) (pg. 20, para. 1 to pg. 21, para. 3 and also pg. 23, para. 2 to pg. 27, para. 2).
This argument is not persuasive. In response to applicant's arguments against the references individually, one cannot show nonobviousness by attacking references individually where the rejections are based on combinations of references. See In re Keller, 642 F.2d 413, 208 USPQ 871 (CCPA 1981); In re Merck & Co., 800 F.2d 1091, 231 USPQ 375 (Fed. Cir. 1986).
First, Harris does disclose a hidden Markov model may be applied to the combined dataset containing both the first and second mapped read sets ([0010], e.g. output generated with a Hidden Markov model; [0043]-[0044]). Simply because Harris describes a particular embodiment in which the hidden Markov model is applied to one sequencing data set does not negate the broader disclosure of Harris. See MPEP 2123 I. A reference may be relied upon for all that it would have reasonably suggested to one having ordinary skill in the art, including nonpreferred embodiments. Merck & Co. v. Biocraft Labs., Inc. 874 F.2d 804, 10 USPQ2d 1843 (Fed. Cir. 1989), cert. denied, 493 U.S. 975 (1989). See also Upsher-Smith Labs. v. Pamlab, LLC, 412 F.3d 1319, 1323, 75 USPQ2d 1213, 1215 (Fed. Cir. 2005).
Regardless, as discussed in the above rejection, Talevich discloses analyzing separate target bins and off-target bins, and converting the bin-level copy ratio estimates into discrete copy number regions using a segmentation algorithm (pg. 7, para. 405). Talevich discloses the off-target bin coverages are generated from untargeted sequencing data while the on-target coverages are determined from targeted reads from panel sequencing (Abstract); Harris discloses untargeted sequencing data can be generated from the non-specific portion of targeted sequencing data (as done in Talevich), or from low coverage whole genome sequencing data ([0009]). In other words, the combination of Harris and Talevich make obvious the use of read coverages from off-target bins from whole-genome sequencing data (i.e. the first mapped dataset) with read coverages of on target bins (i.e. the second mapped dataset) in a segmentation model for copy number variation prediction.
Applicant remarks that while Talevich discloses determining coverages in both off-target bins and on-target bins to determine bin coverages, this does not address the deficiencies in Harris because Talevich applies the segmentation algorithm to the log2 copy ratio values derived from this single combined coverage track, and not to multiple mapped datasets with distinct sequencing contexts, and further remarks that the on-target and off-target bin assignments merely represent different regions of the genome within the same dataset, and do not represent different sequencing modalities, and furthermore, there is no teaching of combining a bin from one mapped dataset with a bin from another mapped dataset into a single segment while preserving original coverage distribution (Applicant’s remarks at pg. 21, para. 4 to pg. 22, para. 1 and pg. 25, para. 2 to pg. 26, para. 1).
This argument is not persuasive. In response to applicant's arguments against the references individually, one cannot show nonobviousness by attacking references individually where the rejections are based on combinations of references. See In re Keller, 642 F.2d 413, 208 USPQ 871 (CCPA 1981); In re Merck & Co., 800 F.2d 1091, 231 USPQ 375 (Fed. Cir. 1986).
As explained above, Talevich discloses analyzing separate target bins and off-target bins, and converting the bin-level copy ratio estimates into discrete copy number regions using a segmentation algorithm (pg. 7, para. 405). Talevich discloses the off-target bin coverages are generated from untargeted sequencing data while the on-target coverages are determined from targeted reads from panel sequencing (Abstract); Harris discloses untargeted sequencing data can be generated from the non-specific portion of targeted sequencing data (as done in Talevich), or from low coverage whole genome sequencing data ([0009]). In other words, the combination of Harris and Talevich make obvious the use of read coverages from off-target bins from whole-genome sequencing data (i.e. the first mapped dataset), as opposed to using off-target reads, with read coverages of on target bins of targeted sequencing data (i.e. the second mapped dataset) in a segmentation model for copy number variation prediction. This corresponds to the two sequencing tracks referred to by Applicant.
Talevich also creates separate mapped datasets from the on-target and off-target reads (Fig. 1). Furthermore, claim 1 only requires that the model is applied to the first and second mapped datasets, which each contain the bin information; however, nothing in the claims precludes the bins from having been merged into a single file (i.e. the log2 ratios of the bins). Talevich discloses the copy ratio estimates are at the bin-level (i.e. copy ratio estimates for off-target bins and on-target bins) (pg. 7, para. 4). Each of the copy ratio estimates of the off-target bins and on-target bins are respectively determined from the off-target bed file and the on-target bed file (Fig. 1, pg. 7, para. 4), and thus the original coverage distributions are preserved, contrary to Applicant’s assertion.
Applicant remarks that the segmentation in Talevich is performed after merging the coverage data and applying bias correction, which is different form the claimed method where segmentation is performed on multi-context data each preserving the original mapping context (Applicant’s remarks at pg. 22, para. 2).
This argument is not persuasive for the reasons discussed above. The log2 ratios of Talevich are at the bin level, with each bin specifically corresponding to an off-target or on-target bin determined from only reads aligning to off-target regions or on-target regions respectively (thus maintaining original coverage distributions, resulting in the assigning as claimed). Furthermore, as discussed above, even if Talevich combined the bin values and/or log2 ratios for the set of off-target bins (i.e. the first mapped dataset) and on-target bins (i.e. the second mapped dataset), this still reads on the claims. Nothing in the independent claim precludes the first and second mapped datasets containing bin values from having been merged into a same file before applying the model to the data.
Applicant remarks that Talevich operates on a per-sample, single-modality coverage track, and there is no disclosure of the claimed segment-level fusion of WGS and targeted data (Applicant’s remarks at pg. 22, para. 3 to 4).
This argument is not persuasive for the reasons already discussed above. In response to applicant's arguments against the references individually, one cannot show nonobviousness by attacking references individually where the rejections are based on combinations of references. See In re Keller, 642 F.2d 413, 208 USPQ 871 (CCPA 1981); In re Merck & Co., 800 F.2d 1091, 231 USPQ 375 (Fed. Cir. 1986).
As explained above, Talevich discloses the off-target bin coverages are generated from untargeted sequencing data while the on-target coverages are determined from targeted reads from panel sequencing (Abstract); Harris discloses untargeted sequencing data can be generated from the non-specific portion of targeted sequencing data (as done in Talevich), or from low coverage whole genome sequencing data ([0009]). In other words, the combination of Harris and Talevich make obvious the use of read coverages from off-target bins from whole-genome sequencing data (i.e. the first mapped dataset), as opposed to using off-target reads, with read coverages of on target bins of targeted sequencing data (i.e. the second mapped dataset) in a segmentation model for copy number variation prediction. This corresponds to the two sequencing tracks referred to by Applicant.
Applicant remarks the dependent claims dependent from claim 1, and thus are not obvious for the same reasons as claim 1 (Applicant’s remarks at pg. 27, para. 3).
This argument is not persuasive for the same reasons discussed above for claim 1.
Applicant remarks that the rejections of claims 18, 28, 37, and 39 should be withdrawn for the same reasons discussed above for claim 1 (Applicant’s remarks at pg. 17, para. 5 to pg. 28, para. 5).
This argument is not persuasive for the same reasons discussed above for claim 1.
Conclusion
No claims are allowed.
Claims 1-4, 6-7, 9-16, 18-19, 24-25, 28, 31, 33-34, and 36-45 are patent eligible for the reasons discussed in the Office action mailed 10 Aug. 2023 and 20 Dec. 2022.
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