Causes and analytical impacts of missing data in RADseq phylogenetics: Insights from an African frog (Afrixalus)

Abstract

Restriction site-associated DNA sequencing (RADseq) has emerged as a useful tool in systematics and population genomics. A common feature of RADseq data sets is that they contain missing data that arise from multiple sources including genealogical sampling bias, assembly methodology and sequencing error. Many RADseq studies have demonstrated that allowing sites (single nucleotide polymorphisms, SNPs) with missing data can increase support for phylogenetic hypotheses. Two non-mutually exclusive explanations for this observation are that (a) larger data sets contain more phylogenetic information; and (b) excluding missing data disproportionally removes sites with the highest mutation rates, causing the exclusion of characters that are likely variable and informative. Using a RADseq data set derived from the East African banana frog, Afrixalus fornasini (up to 1.1 million SNPs), we found that missing data thresholds were positively correlated with the proportion of parsimonyinformative sites and mean branch support. Using three proxies for estimating sitespecific rate, we found that the most conservative missing data strategies excluded rapidly evolving sites, with four-state sites present only when allowing ≥60% missing data per SNP. Topological similarity among estimated phylogenies was highest for the data sets with ≥60% missing data per SNP. Our results suggest that several desirable phylogenetic qualities were observed when allowing ≥60% missing data per SNP. However, at the highest missing data thresholds (80% and 90% missing data per SNP), we observed differences in performance between high-and mixedweight DNA extraction samples, which may indicate there are trade-offs to consider when using degraded genomic template with RADseq protocols. |CROTTI eT al.

Performance comparison of two reduced-representation based genome-wide marker-discovery strategies in a multi-taxon phylogeographic framework

“…A correlation between sample size (i.e. number of sampled populations per region) and phylogenetic resolution has been demonstrated in other RADseq-based phylogenetic studies 47 , 48 and for several AFLP-inferred population genetic measures 49 . Small sample sizes and missing data might not be as much a problem for RADseq as for AFLP.…”

Incomplete lineage sorting and ancient admixture, and speciation without morphological change in ghost-worm cryptic species

“…Missing data is especially problematic in RADseq as it can lead to erroneous inference of population-genetic parameters ( Arnold et al, 2013 ; Gautier et al, 2013 ; Hodel et al, 2017 ). However, applying stringent filtering for missing data has been shown to prune parsimonious-informative loci, with best-practices suggesting non-conservative pruning of the data ( Huang & Lacey Knowles, 2016 ; Lee et al, 2018 ; Crotti et al, 2019 ). To mitigate missing data while avoiding stringent filtering, we applied a novel procedure, which allowed us to retrieve more loci from our data ( Cerca et al, 2021 ).…”

Removing the bad apples: A simple bioinformatic method to improve loci‐recovery in de novo RADseq data for non‐model organisms

“…Rates of allele dropout are thereby expected to be correlated with the divergence between lineages (Crotti et al., 2019; Eaton et al., 2017; O'Leary et al., 2018). However, allele dropout may also result from artefacts in the experimental design, such as sampling bias and low sequence coverage; or from problems associated with library preparation, such as issues with enzyme digestion or size selection, and human error; or from challenges in DNA extraction since, for some organismal groups, extracting DNA may still be non‐trivial due to their reduced size or presence of chemical compounds which may interfere with the extraction; or from artefacts from bioinformatic analyses, such as problems associated with clustering of sequencing reads (Crotti et al., 2019; O'Leary et al., 2018). In fact, allele dropout originating from these technical artefacts can sometimes exceed dropout of biological origin under certain experimental conditions (Rivera‐Colón et al., 2020).…”

“…In fact, allele dropout originating from these technical artefacts can sometimes exceed dropout of biological origin under certain experimental conditions (Rivera‐Colón et al., 2020). Whatever the case may be, high allele dropout translates to high rates of missing data in the dataset, which may dramatically influence allele frequency in the dataset (Arnold et al., 2013; Gautier et al., 2013; Hodel et al., 2017), or phylogenetic reconstruction (Crotti et al., 2019; Eaton et al., 2017).…”

“…Attempts to minimize the challenges posed by allele dropout include suggestions for parameter optimization and control (Paris et al., 2017; Rochette & Catchen, 2017), data‐filtering and data exploration (O'Leary et al., 2018), data‐cleaning thresholds (Crotti et al., 2019), and prospective and retrospective data simulation based on the reference genome available (Rivera‐Colón et al., 2020). Despite these, retrieving an optimal dataset from a RADseq experiment can pose challenges.…”