Single-Sample Germline Variant Filtering for Somatic Calling
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing methods for variant calling from sequence data struggle to accurately distinguish somatic variants from germline variants without requiring a matched normal sample, leading to potential overestimation of tumor mutation burden.
Innovation Solution
A method involving a database filter and a proximity filter is applied to sequence data from a single sample to identify and remove germline variants based on allele count and genomic location, allowing for the distinction between somatic and germline variants.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Device complexity
If variant calling is performed without a matched normal sample, then the complexity of sample requirements is reduced, but the accuracy of distinguishing somatic variants from germline variants deteriorates
Solution Approach 1:
The patent introduces a germline variant database as an intermediary resource to mediate the classification process. This database contains pre-identified germline variants from population studies, which serve as a reference to filter out germline variants from tumor sequencing data without requiring a matched normal sample. The database acts as a mediator between the tumor sample and the variant classification process, enabling accurate somatic variant identification while simplifying sample requirements.
Solution Approach 2:
The patent applies preliminary filtering actions by first removing variants that are known to be germline based on database matches before performing somatic variant calling. This preliminary action of filtering out germline variants using allele frequency thresholds and database comparisons prepares the data for more accurate somatic variant identification, effectively simulating the role of a matched normal sample through pre-computed population data.
2Productivity
If traditional variant calling methods are used without germline filtering, then the processing speed is maintained, but the reliability of tumor mutation burden estimation deteriorates
Solution Approach 1:
The patent implements preliminary germline variant filtering using database comparisons and allele frequency thresholds before final variant calling. This preliminary action removes known germline variants from the candidate list, ensuring that only potential somatic variants proceed to detailed analysis. This approach maintains processing speed by using efficient database lookups and threshold-based filtering, while simultaneously improving the reliability of tumor mutation burden estimation by eliminating false positive germline variants.
Solution Approach 2:
The patent changes key parameters for variant evaluation by introducing allele frequency thresholds and database match criteria as additional filtering parameters. Instead of relying solely on traditional variant calling metrics, the method incorporates allele frequency from population databases and proximity to known germline variants as new parameters. These parameter changes enable rapid filtering of germline variants while maintaining high sensitivity for somatic variant detection, thus preserving productivity while enhancing reliability.
Data Source
AI summary
Some embodiments of the methods and systems provided herein relate to variant calling from sequence data obtained from a single sample. In some embodiments, a somatic variant can be distinguished from a germline variant based on variant allele frequency in a sample and location in a genome.