Population-Specific Genomic Graphs for Accurate TMB Estimation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional methods for determining tumor mutational burden (TMB) are unreliable due to misclassification of germline variants as somatic variants, especially in non-European populations, leading to inaccurate TMB estimates and therapeutic mispredictions.
Innovation Solution
Utilizing a population-specific genomic reference graph to align sequence reads from a tumor sample, identifying somatic variants, and determining TMB, which accurately distinguishes between somatic and germline variants without requiring additional non-tumor samples.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If conventional methods are used to determine TMB, then the process is simple, but the accuracy of TMB estimation deteriorates due to misclassification of germline variants as somatic variants
Solution Approach 1:
The reference genome is segmented into population-specific components by creating separate reference genomes for different ancestral populations (e.g., European, African, Asian). This segmentation allows the method to account for population-specific germline variants without requiring a single complex comprehensive reference, thereby improving TMB estimation accuracy while managing complexity through modular population-specific references.
Solution Approach 2:
A graph-based variant representation system serves as an intermediary between the sequence reads and the reference genome. This intermediary structure efficiently represents both the linear reference sequence and population-specific variants, enabling accurate distinction between germline and somatic variants without directly complicating the alignment process. The graph structure acts as a mediator that integrates multiple population-specific variations in a computationally manageable way.
2Measurement precision
If additional non-tumor samples are required to distinguish germline variants, then variant classification accuracy improves, but the complexity and cost of the procedure increases
Solution Approach 1:
Population-specific germline variants are identified and incorporated into the reference genome in advance, before the actual TMB analysis of tumor samples. This preliminary action creates a pre-enriched reference that already contains knowledge about common germline variants for each population, allowing direct distinction between germline and somatic variants in tumor samples without requiring additional non-tumor sample collection or analysis.
Solution Approach 2:
Instead of requiring actual non-tumor samples from each patient, the method creates a computational copy of population-specific germline variant information from large cohorts and integrates it into the reference genome. This copying approach allows the system to simulate the effect of having patient-specific non-tumor samples without the practical burden of collecting and processing additional biological samples.
3Measurement precision
If population-specific variants are incorporated into the reference, then TMB determination accuracy for diverse populations improves, but the reference genome size increases
Solution Approach 1:
The reference genome is customized with population-specific variants only in the local regions where such variants are prevalent, rather than uniformly across the entire genome. This local quality approach ensures that non-European populations benefit from accurate representation of their specific variants while avoiding the unnecessary inclusion of rare or population-specific variants in other genomic regions, thus managing reference genome size efficiently.
Solution Approach 2:
The reference genome structure is made dynamic and adaptable to different population contexts. Rather than creating entirely separate large reference genomes for each population, the system dynamically adjusts the reference by incorporating population-specific variants where needed and using the standard reference where population-specific information is unavailable, optimizing the balance between accuracy and data volume based on the specific population being analyzed.
Data Source
AI summary
Described herein are techniques for determining tumor mutational burden (TMB) of a tumor sample previously obtained from a subject. In some embodiments, the techniques include obtaining sequence reads, the sequence reads having been previously obtained by sequencing the tumor sample; aligning the sequence reads to a population-specific genomic reference graph representing a linear reference sequence and population-specific variants relative to the linear reference sequence, wherein the population-specific variants are variants associated with at least one population to which the subject belongs; identifying, based on a result of aligning the sequence reads to the population-specific genomic reference graph, a plurality of somatic variants; and determining the TMB of the tumor sample using the identified plurality of somatic variants.