SLAFIT: Viral fitness inference and structural variant detection from TGS data

Evolution is driven by genetic variation at multiple scales, from single-nucleotide variants (SNVs) to structural variants (SVs). Within populations, nonrandom associations among SNVs, quantified by linkage disequilibrium (LD), must be considered when inferring fitness from sequencing data. Characterizing SVs, SNVs, and the LD between SNVs therefore provides complementary insights into the genetic basis of evolutionary change. Although third-generation sequencing technologies facilitate the characterization of genomic variation, finite read lengths often limit the resolution of LD across entire genomes. Here, we present SLAFIT, a pipeline for linkage-aware fitness inference and SV detection from third-generation sequencing data. The pipeline reconstructs genome-wide haplotypes and their frequencies from third-generation sequencing reads to infer SNVs and estimate LD. When time-series data are available, it also infers fitness. An established SV detection tool is incorporated into the same workflow. Simulations support accurate LD and fitness inference under tested conditions, with reduced computation time through parallelization. Applications to enterovirus C104 and herpes simplex virus 1 reveal selection signals in predicted epitopes and recover a known inversion, respectively. Together, these results demonstrate the utility of SLAFIT for studying viral evolution through complementary analyses of fitness and structural variation.