SNV-PPILP

SNV-PPILP refines single-nucleotide variant (SNV) calls from GATK's Unified Genotyper using phylogenetic information to improve variant-call accuracy across multiple samples.

Key Features:

• Phylogeny-Based Refinement: Incorporates phylogenetic relationships among samples to adjust SNV calls, benefiting analyses such as tumor samples collected across developmental stages.
• Improved Low-Coverage Performance: Demonstrates higher accuracy than GATK's Unified Genotyper, particularly in low read coverage scenarios.
• Implementation and Solver: Implemented in Python and requires the ILP solver lp_solve to perform the optimization.
• Validation on Simulations: Tested on simulated datasets varying numbers of samples, SNVs, read coverage, and violations of the perfect phylogeny assumption, matching or surpassing GATK's Unified Genotyper.

Scientific Applications:

• Cancer Genomics: Refines SNV calls to support reconstruction of tumor evolutionary histories across multiple samples.
• Early-Stage Tumor Detection: Improves detection of early-stage tumors via more accurate SNV identification.
• Understanding Cancer Progression: Clarifies the distribution and temporal order of genetic variants to inform studies of tumor progression.
• Drug Targeting and Precision Medicine: Provides more accurate SNV profiles to inform drug target identification and treatment stratification.

Methodology:

Refines initial SNV calls produced by GATK's Unified Genotyper by imposing a phylogeny-based model, implemented in Python and solved as an integer linear program using lp_solve, and designed to handle varying read coverages and deviations from a perfect phylogeny.