Comparative study of population genomic approaches for mapping colony-level traits

Social insect colonies exhibit colony-level phenotypes such as social immunity and task coordination, which are produced by the individual phenotypes. Mapping the genetic basis of such phenotypes requires associating the colony-level phenotype with the genotypes in the colony. In this paper, we examine alternative approaches to DNA extraction, library construction, and sequencing for genome wide association studies (GWAS) of colony-level traits using a population sample of Cataglyphis niger ants. We evaluate the accuracy of allele frequency estimation from sequencing a pool of individuals (pool-seq) from each colony using either whole-genome sequencing or reduced representation genomic sequencing. Based on empirical measurement of the experimental noise in sequenced DNA pools, we show that reduced representation pool-seq is drastically less accurate than whole-genome pool-seq. Surprisingly, normalized pooling of samples did not result in greater accuracy than un-normalized pooling. Subsequently, we evaluate the power of the alternative approaches for detecting quantitative trait loci (QTL) of colony-level traits by using simulations that account for an environmental effect on the phenotype. Our results can inform experimental designs and enable optimizing the power of GWAS depending on budget, availability of samples and research goals. We conclude that for a given budget, sequencing un-normalized pools of individuals from each colony provides optimal QTL detection power.
Author summary Genomic mapping techniques are used to map phenotypes to genotypes. Mapping is of general interest in any biological system, including fundamental studies of biological traits, clinical studies of genetic predisposition to disease, and agro- and bio-technological studies of domesticated plants and animals. Typically, such studies associate phenotypic measurements of individuals with their genotypes. Here we evaluate methodological approaches for genomic mapping of phenotypes that are expressed at the level of a group rather than that of individuals. We demonstrate that genomic sequencing of a DNA pool from multiple samples provides increased statistical power within a limited budget. Our results facilitate more efficient use of resources in genomic mapping studies that investigate group-level phenotypes.