|
Abstract:
|
In microbial forensics, assessing whether sample A and B are from the same source remains difficult. This problem is particularly challenging for clonal bacterial pathogens with low genetic variability, such as B. anthracis and F. tularensis - both Category A Priority pathogens. Ultra-deep whole genome shotgun sequencing can, in theory, provide sample attribution by focusing on minority and rare genetic variants (SNPs). Such deep sequencing however, is prone to significant SNP false discovery rates due to errors in every step of the process: from sample preparation, to sequence generation and alignment, to SNP detection. These errors greatly reduce variant detection specificity, and thus utility, of the entire approach. Herein, we present our efforts to address these challenges by developing exhaustive sequence alignment and associated SNP detection algorithms, as well as rare SNP profile-based sample comparison statistical frameworks. As a consequence, we have been able to reduce SNP false positive rates by an order of magnitude, achieving sensitive and specific discrimination between samples that differ by admixtures of rare variant strains present in as little as 0.1%.
|
