Origin matters: Using a local reference genome improves measures in population genomics

Abstract

Genome sequencing enables answering fundamental questions about the genetic basis of adaptation, population structure and epigenetic mechanisms. Yet, we usually need a suitable reference genome for mapping population‐level resequencing data. In some model systems, multiple reference genomes are available, giving the challenging task of determining which reference genome best suits the data. Here, we compared the use of two different reference genomes for the three‐spined stickleback (Gasterosteus aculeatus), one novel genome derived from a European gynogenetic individual and the published reference genome of a North American individual. Specifically, we investigated the impact of using a local reference versus one generated from a distinct lineage on several common population genomics analyses. Through mapping genome resequencing data of 60 sticklebacks from across Europe and North America, we demonstrate that genetic distance among samples and the reference genomes impacts downstream analyses. Using a local reference genome increased mapping efficiency and genotyping accuracy, effectively retaining more and better data. Despite comparable distributions of the metrics generated across the genome using SNP data (i.e. π, Tajima's D and FST), window‐based statistics using different references resulted in different outlier genes and enriched gene functions. A marker‐based analysis of DNA methylation distributions had a comparably high overlap in outlier genes and functions, yet with distinct differences depending on the reference genome. Overall, our results highlight how using a local reference genome decreases reference bias to increase confidence in downstream analyses of the data. Such results have significant implications in all reference‐genome‐based population genomic analyses.

Demographic history and genomic signatures of selection in a widespread vertebrate ectotherm

“…Further, the number of short ROH increased, consistent with bottlenecks leading to the current low levels of genetic variation in Oregon and Idaho, and to a lesser extent in the New Mexico population. ROH analysis with RADseq data is only an approximation of ROH dynamics across the genome, and the reference we used (from the northwestern US) may have slightly biased our ROH to be lower in eastern populations (by inflating heterozygosity, Thorburn et al., 2023). However, the number of markers used and the agreement with demographic results, which have been found to be less sensitive to reference genetic distance when divergence is less than 3% between reference and target species (Prasad et al., 2022), suggest that the increasing number of short ROH from eastern to western populations is likely a close representation of ROH in western painted turtles (Shafer et al., 2016).…”

Population genomics of harbour seal Phoca vitulina from northern British Columbia through California and comparison to the Atlantic subspecies

“…Further, we evaluate and find that this trend is not caused by reference genome bias in genotyping, where individuals with increased divergence from the reference genome may exhibit biased population genetic metrics (Bohling, 2020). Reference bias is typically considered in the context of divergent species, but as more local genomes are produced, these issues may become more relevant within species (Beichman et al, 2023;Thorburn et al, 2023). Further, given the high divergence between the subspecies here, it was important to test whether P. The ultimate cause of the diversity difference was proposed by Liu et al (2022) to be due to stepwise founder effects during colonization from the Pacific Ocean to the western Atlantic Ocean, and then from the western Atlantic Ocean to the eastern Atlantic Ocean, with a decrease in genetic diversity at each step.…”

Genomics and conservation: Guidance from training to analyses and applications

“…This is because the differences in the sequence order and content between species can ultimately affect downstream analyses, a consequence broadly termed 'reference bias' (Günther & Nettelblad, 2019;Prasad et al, 2022). In fact, reference bias can even occur within species (Thorburn et al, 2023). To accommodate within-species reference bias, reference genomes from several individuals within a species can also be used to construct pangenomes, which further improves our ability to represent diversity within a species (e.g.…”