Fast genomic analysis of aquatic bird populations from short single-end reads considering sex-related pitfalls

摘要

Population genomics is transforming the analyses of biodiversity, particularly the studies in conservation genetics. However, genomic approaches based on short sequencing reads are often subjected to complex molecular protocols and bioinformatics procedures, which are prone to many artifactual biases. In preliminary analyses using common pipelines for handling, filtering and processing genotyping-by-sequencing (GBS) data from three aquatic bird species, we detected a bias likely associated to the sex of individuals of the genus Sula, which could be due to artifacts with sex chromosome derived reads. Thus, we have developed GBirdS, a new pipeline for avian population genetics from GBS data, implementing a new approach to identify molecular markers likely to be linked to the avian W chromosome. The pipeline was applied to a total of 113 samples from three avian species (Sula leucogaster, Sula dactylatra and Mergus octoseraces), including 54 samples for which the molecular sexing was available. For 53 out of these 54 samples, the sex determination achieved using GBirdS agreed with molecular sexing. We have then removed from subsequent population analyses the sequences observed only in individuals determined as heterogametic (likely to be females) by our pipeline. Removing these sequences resulted in greater average fixation indexes between subpopulations of these three species. Therefore, we conclude that the sequences removed by our pipeline had an artificial effect on interindividual differences and population structure.