FRAGMATIC: in silico locus prediction and its utility in optimizing ddRADseq projects

Abstract

Reduced-representation genomic methods are an invaluable data acquisition tool for conservation geneticists, yet a priori estimates of locus recovery are difficult for non-model organisms. We present a simple in silico approach (FRAGMATIC) that predicts locus recovery in ddRAD sequencing which utilizes genomic data for related organisms. Its applicability was tested by quantifying prediction accuracy versus genetic distances across five non-model organisms and reference genomes for related organisms of varying phylogenetic distance. We additionally examined sensitivity of the method using one organism (Danio rerio) with an available genome. FRAGMATIC supports population genomic projects in non-model species by providing a priori estimates of targeted ddRAD loci that, in turn, will curb wasted sequencing effort and optimize cost-efficiency. Validation shows that while predictive error is minimized when applied to a closely related reference genome, in silico estimates may also be robust to deeper (e.g. within-family) relationships, although weak correlation suggests that specific characteristics of genome architecture may be more predictive than genetic distance. This indicates that a more extensive exploration of genomes, including a broader taxonomic scope (e.g. beyond vertebrates), may be informative. All code is freely available at: https://github.com/tkchafin/fragmatic.