The impact of adaptor selection on genotyping in 2b-RAD studies

Abstract

Population genomic studies have increased in the last decade, showing great potential to understand the evolutionary patterns in a great variety of organisms, mostly relying on RAD sequencing techniques to obtain reduced representations of the genomes. Among them, 2b-RAD can provide further secondary reduction to adjust experimental costs by using base-selective adaptors, although its impact on genotyping is unknown. Here we provide empirical comparisons on genotyping and genetic differentiation when using fully degenerate or base-selective adaptors and assess the impact of missing data. We built libraries with the two types of adaptors for the same individuals and generated independent and combined datasets with different missingness filters (presence in 100%, 75% and 50% of the libraries). Our approach, by exploring locus-by-locus, found 92% of identical genotypes between the two libraries of the same individual when using loci shared among 100% of the libraries, which decreased to 35% when working with loci present in at least 50% of them. We show that missing data is a major source of individual genetic differentiation. The loci discordant by genotyping were in low frequency (7.67%) in all filtered files. Only 0.96% of them were directly attributable to base-selective adaptors reducing heterozygosity. An underestimation of heterozygosity in 6.44% of genotypes was found in libraries generated with fully degenerated adaptors, of which ca. 70% had <10 reads per locus indicating that sufficient read depth should be ensured for a correct genotyping. Base-selective adaptors reduce the number of loci, increase their coverage, and provide a reliable genotyping with no heterozygosity underestimation at a lower sequencing cost. We demonstrate that missing data is a major concern for a correct identification of genetic differentiation, and should not be neglected in genomic studies. Thus, we conclude that 2b-RAD libraries using base-selective adaptors are a robust tool to use in population genomics regardless the target species genome size, promoting their use in future studies.