Methods for detecting variants in next- generation sequencing genomic data
Inventors
Hernandez, David • XU, M. Zhenyu
Assignees
Interested in licensing this patent?
MTEC can help explore whether this patent might be available for licensing for your application.
Publication Number
US-12573472-B2
Publication Date
2026-03-10
Expiration Date
Abstract
A genomic data analyzer workflow may be configured to identify, with a variant annotation module, subsets of patient variants which match at least one medical reference variant database entry, even if the variant calling information in genomic data analyzer workflow and the database use different variant representations of SNP, MNP, INDELS and DELINS. In particular, database variants which are included into a subset of patient variants may be identified even if they do not exactly match the corresponding strings. The variant annotation module may be adapted to apply a branch-and-bound-like algorithm to efficiently process all possible subsets of patient variants in a genomic region.
Core Innovation
The invention relates to identifying, with a variant annotation module on at least one data processor, whether variant calling information of a patient sample includes at least one variant entry in a plurality of medical variant reference databases. Variant calling information is acquired from next generation sequencing data by comparing the next generation sequencing data to a reference genome to retrieve a set V* of genomic variants in at least one genomic region to be analyzed.
For each reference variant database entry d in a retrieved subset D, the variant annotation module reconstructs, over an extended genomic region coordinate range a . . . b, a database variant haplotype string h_a...b(d) as the concatenation of RP_left + alt(d) + RP_right. In parallel, over the same extended coordinate range a . . . b, the variant annotation module reconstructs a patient variant haplotype string h_a...b(X) for a subset X of selected patient variants from V.
A database entry d is identified as included in the patient variant calling information when the length of h_a...b(d) equals the length of h_a...b(X), the reconstructed strings differ by at least one mismatch in either the RP_left string or the RP_right string, and every variant in the subset X overlaps by at least one symbol with the location defined by the start and end coordinates of the database variant entry d. The search is performed sequentially across the plurality of medical variant reference databases, and candidate search spaces are constrained by selecting an extended genomic region coordinate range and using global and local horizon parameters to restrict candidate reference variants and nearby patient variants.
The invention further enables efficient exploration of patient variant subsets and haplotype combinations using a binary tree representation, in which subsets X⊆V are explored for matching reconstruction. Branches corresponding to non-matching branches are pruned while matching branches are constructed to generate h_a...b(X) consistent with the reconstructed database entry haplotype string h_a...b(d). In addition, phasing information can be used so that a patient haplotype is identified only when variants in the subset X are in-phase, with optional post-processing confirming composed haplotypes when in-phase.
Claims Coverage
The independent claim is clm-00001, which sets out the overall variant annotation and inclusion decision framework across multiple medical variant reference databases. The main inventive features include extended-range haplotype reconstruction and an inclusion rule that combines reconstructed length equality, flanking-context mismatches, and overlap of patient variants with the database entry coordinates, with the search constrained by horizons and optionally refined by binary-tree subset exploration and phasing-based confirmation.
Extended-range haplotype replay for database entries
Reconstructing, over an extended genomic region coordinate range a . . . b, a database variant haplotype string h_a...b(d) as RP_left + alt(d) + RP_right, where RP_left and RP_right correspond to extended genomic region on the left and right of alt(d) in the reference human genome R.
Extended-range haplotype replay for patient variant subsets
Reconstructing, over the extended genomic region coordinate range a . . . b, a patient variant haplotype string h_a...b(X), where X is a subset of selected region variant calling information V comprising at least one selected patient variant v_j.
Inclusion rule based on length equality, flanking-context mismatch, and overlap
Identifying that the patient variant calling information includes the medical variant database entry d if the length of h_a...b(d) equals the length of h_a...b(X), the reconstructed haplotype strings differ by at least one mismatch in either the RP_left string or the RP_right string, and every variant in subset X overlaps by at least one symbol with the location defined by the start and end coordinates of database variant entry d.
Sequential search across multiple medical variant reference databases
Sequentially searching the plurality of medical variant reference databases for all reference variant database entries as part of the variant annotation module process.
Horizon-constrained candidate selection for reference and patient variants
Selecting a subset D of reference variant database entries that match at least one patient genomic variant V* according to a horizon parameter z, and selecting, for each reference variant database entry d in D, a subset V of patient variants using a local horizon parameter z_d based on start/end coordinate proximity measured in symbols.
Binary-tree subset exploration for matching haplotype reconstruction
Representing possible haplotype combinations derived from patient variants as a binary tree of subsets X⊆V, including pruning of non-matching branches by constructing matching branches (or including-d branches) and aborting construction under non-matching branches.
Phasing-based confirmation of composed haplotypes
Comparing phasing information across a subset X of at least two patient variants and identifying a patient haplotype from those variants only when all involved variants are in-phase.
The claim coverage centers on reconstructing database and patient haplotype strings over an extended coordinate range and using an inclusion rule requiring reconstructed length equality, flanking-context mismatch in RP_left or RP_right, and overlap of each patient variant in the subset with the database entry coordinates. Dependent claim themes add horizon-constrained candidate selection, binary-tree subset exploration with pruning, and optional phasing-based confirmation requiring in-phase variants.
Stated Advantages
Allows identifying whether patient variant calling information includes at least one variant entry in a plurality of medical variant reference databases.
Enables efficient search by constraining candidate database entries and candidate patient variants using horizon parameters.
Improves subset/haplotype reconstruction efficiency by using binary-tree exploration with pruning of non-matching branches.
Supports haplotype confirmation using phasing information by requiring variants to be in-phase.
Documented Applications
Variant annotation workflow context including alignment, variant calling, and variant annotation across medical reference databases such as ClinVar and COSMIC.
Experimental examples reporting included and composed confirmed matches in BRCA2 (rs80358464, rs80359642) using ClinVar and in ARSD (COSM5421975) using COSMIC.
Interested in licensing this patent?