Bioinformatic processing of RAD‐seq data dramatically impacts downstream population genetic inference

Abstract

Summary Restriction site‐associated DNA sequencing (RAD‐seq) provides high‐resolution population genomic data at low cost, and has become an important component in ecological and evolutionary studies. As with all high‐throughput technologies, analytic strategies require critical validation to ensure precise and unbiased interpretation. To test the impact of bioinformatic data processing on downstream population genetic inferences, we analysed mammalian RAD‐seq data (>100 individuals) with 312 combinations of methodology (de novo vs. mapping to references of increasing divergence) and filtering criteria (missing data, HWE, FIS, coverage, mapping and genotype quality). In an effort to identify commonalities and biases in all pipelines, we computed summary statistics (nr. loci, nr. SNP, π, Hetobs, FIS, FST, Ne and m) and compared the results to independent null expectations (isolation‐by‐distance correlation, expected transition‐to‐transversion ratio Ts/Tv and Mendelian mismatch rates of known parent–offspring trios). We observed large differences between reference‐based and de novo approaches, the former generally calling more SNPs and reducing FIS and Ts/Tv. Data completion levels showed little impact on most summary statistics, and FST estimates were robust across all pipelines. The site frequency spectrum was highly sensitive to the chosen approach as reflected in large variance of parameter estimates across demographic scenarios (single‐population bottlenecks and isolation‐with‐migration model). Null expectations were best met by reference‐based approaches, although contingent on the specific criteria. We recommend that RAD‐seq studies employ reference‐based approaches to a closely related genome, and due to the high stochasticity associated with the pipeline advocate the use of multiple pipelines to ensure robust population genetic and demographic inferences.

Phylogenomic Insights into Mouse Evolution Using a Pseudoreference Approach

“…Recent studies using restriction-site associated DNA sequencing (RAD-seq) have shown that overall data quality is considerably higher when using a reference (Fountain et al. 2016; Shafer et al. 2016).…”

Comparing Pool‐seq, Rapture, and GBS genotyping for inferring weak population structure: The American lobster (Homarus americanus) as a case study

“…Nevertheless, allele frequency estimates remained similar between individual-based methods and Pool-seq, except for one pool. Moreover, as stipulated by Shafer et al (2017), it is difficult to reproduce the important variation introduced during wet laboratory data generation using simulated data. (a, b, and c) represent the eigenvalue decomposition of the scaled variance-covariance matrices of population allele frequencies (Ω) for GBS, Rapture, and Pool-seq datasets, respectively.…”

Recent advances in conservation and population genomics data analysis

“…Furthermore, filtering choices (see figure 2 in Benestan et al., 2016) can greatly influence downstream summary statistics. A recent study testing the impact of data processing on population genetic inferences using RAD‐seq data observed large differences between reference‐based and de novo approaches in population genetic summary statistics, particularly those based on the site frequency spectrum (Shafer et al., 2016). In addition, the recent debate over the effectiveness of RAD‐seq for discovering loci under selection (Catchen et al., 2017; Lowry et al., 2016; McKinney, Larson, Seeb, & Seeb, 2017) has highlighted the importance of testing the extent of linkage disequilibrium (LD) over the genome, whenever possible, in order to assess the power of genome scans to detect selected loci (e.g., Kardos, Taylor, Ellegren, Luikart, & Allendorf, 2016).…”

“…On the other hand, imposing MAF filters that are too strict (e.g., above 0.05 or 0.1) could skew metrics based on the site frequency spectrum or inadvertently remove loci under selection or with functional significance. As others have recommended, testing the effects of a range of analytical (filtering) parameters is critical to produce robust population genetic and demographic inferences (Mastretta‐Yanes et al., 2014; Paris, Stevens, & Catchen, 2017; Shafer et al., 2016). …”

“…Despite this, having even a draft genome (in 1000s of scaffolds) can help improve data analyses in many ways including the following: (a) reliable discovery of SNPs (e.g., avoiding duplicated loci), (b) reducing genotyping error rates (Hand et al., 2015; Shafer et al., 2016), (c) detecting loci under selection by allowing sliding‐window approaches along scaffolds (Hohenlohe, Phillips, & Cresko, 2010), and (d) finding the underlying genes associated with phenotype or adaptation (facilitated by mapping scaffolds to related species with well‐annotated genomes; e.g., Ekblom & Wolf, 2014; Kohn, Murphy, Ostrander, & Wayne, 2006; McKinney et al., 2016). In addition, it is possible with this information to estimate effective population size ( N e ; e.g., Li & Durbin, 2011) or effective number of breeders ( N b ) using LD‐based methods, as comparisons can be restricted to pairs of loci on different scaffolds, which should reduce or eliminate LD due to physical linkage.…”