Simulation with RADinitio Improves RADseq Experimental Design and Sheds Light on Sources of Missing Data

Abstract

<jats:title>Abstract</jats:title><jats:p>Restriction-site Associated DNA sequencing (RADseq) has become a powerful and versatile tool in modern population genomics, enabling large-scale genomic analyses in otherwise inaccessible biological systems. With its widespread use, different variants on the protocol have been developed to suit specific experimental needs. Researchers face the challenge of choosing the optimal molecular and sequencing protocols for their experimental design, an often-complicated process. Strategic errors can lead to improper data generation that has reduced power to answer biological questions. Here we present <jats:italic>RADinitio</jats:italic>, simulation software for the selection and optimization of RADseq experiments via the generation of sequencing data that behaves similarly to empirical sources. <jats:italic>RADinitio</jats:italic> provides an evolutionary simulation of populations, implementation of various RADseq protocols with customizable parameters, and thorough assessment of missing data. Using the software, we test its efficacy using different RAD protocols across several organisms, highlighting the importance of protocol selection on the magnitude and quality of data acquired. Additionally, we test the effects of RAD library preparation and sequencing on allelic dropout, observing that library preparation and sequencing often contributes more to missing alleles than population-level variation.</jats:p>

Stacks 2: Analytical methods for paired‐end sequencing improve RADseq‐based population genomics

“…The number of SNP and genotype calls may be inflated by erroneous calls, and thus cannot in itself be regarded as an indicator of the quality of these inferences. In order to assess the quality of the Stacks v2 SNP and genotype calls and their spatial variations within loci, we created simulation data sets with the radinitio package (Rivera‐Colón, Rochette, & Catchen ), using the threespine stickleback reference genome as template (see File ). We simulated two data sets, one high coverage (20×) and one low coverage (10×), each comprising 100 individuals split among four populations.…”

Customize and get the most out of your reduced‐representation sequencing experiment with the new simulation software RADinitio

“…The article by Rivera‐Colón et al (2020) in the current issue of Molecular Ecology Resources introduces the new simulation software RADinitio . The function of RADinitio is to simulate behaviours of different RADseq protocols and parameters of library preparation on a particular species in generating sequencing data.…”

“…Rivera‐Colón et al (2020) describe RADinitio as a three‐step pipeline. The user first needs to feed the software with a reference sequence that will be used to simulate a metapopulation.…”

“…The level of detail thought by the authors to make RADinitio mimic the effects of several variables (via parameters defined by the user) on data generated by a RADseq experiment brings RADseq simulation software to a new level. By considering allele dropout resulting from natural polymorphism in RADinitio simulations on a set of species, Rivera‐Colón et al (2020) were able to characterize for the first time the degree of contribution of library preparation and sequencing parameters to total allele dropout. They found that using a bad template quality and a suboptimal sequencing coverage influence much more the amount of allele dropout compared to the effect of natural polymorphism at enzyme restriction sites.…”

Editorial 2021

“…Top resources published in Molecular Ecology Resources in 2020 include key contributions across multiple subject areas of the journal. Three separate but complementary computer programs, which are at the frontier of their respective fields, were published: RADinitio, which allows users to simulate RAD (restriction site‐associated DNA) data to aid experimental design (Rivera‐Colón et al, 2020); Jackalope, which enables users to generate high‐throughput sequencing simulations that can take complex evolutionary scenarios into account (Nell, 2020); and a new module of the landscape genetics programs CDPOP and CDMetaPOP, which enables users to simulate multilocus adaption in complex environments in order to better evaluate genotype–environment associations (Landguth et al., 2020). To ensure that all researchers have access to the latest tools in their field, computer programs continue to be published as ‘free access’.…”