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

For half a century population genetics studies have put type II restriction endonucleases to work. Now, coupled with massively‐parallel, short‐read sequencing, the family of RAD protocols that wields these enzymes has generated vast genetic knowledge from the natural world. Here, we describe the first software natively capable of using paired‐end sequencing to derive short contigs from de novo RAD data. Stacks version 2 employs a de Bruijn graph assembler to build and connect contigs from forward and reverse reads for each de novo RAD locus, which it then uses as a reference for read alignments. The new architecture allows all the individuals in a metapopulation to be considered at the same time as each RAD locus is processed. This enables a Bayesian genotype caller to provide precise SNPs, and a robust algorithm to phase those SNPs into long haplotypes, generating RAD loci that are 400–800 bp in length. To prove its recall and precision, we tested the software with simulated data and compared reference‐aligned and de novo analyses of three empirical data sets. Our study shows that the latest version of Stacks is highly accurate and outperforms other software in assembling and genotyping paired‐end de novo data sets.     

Resolving allele dosage in duplicated loci using genotyping‐by‐sequencing data: A path forward for population genetic analysis

Whole‐genome duplications have occurred in the recent ancestors of many plants, fish and amphibians. Signals of these whole‐genome duplications still exist in the form of paralogous loci. Recent advances have allowed reliable identification of paralogs in genotyping‐by‐sequencing (GBS) data such as that generated from restriction‐site‐associated DNA sequencing (RADSeq); however, excluding paralogs from analyses is still routine due to difficulties in genotyping. This exclusion of paralogs may filter a large fraction of loci, including loci that may be adaptively important or informative for population genetic analyses. We present a maximum‐likelihood method for inferring allele dosage in paralogs and assess its accuracy using simulated GBS, empirical RADSeq and amplicon sequencing data from Chinook salmon. We accurately infer allele dosage for some paralogs from a RADSeq data set and show how accuracy is dependent upon both read depth and allele frequency. The amplicon sequencing data set, using RADSeq‐derived markers, achieved sufficient depth to infer allele dosage for all paralogs. This study demonstrates that RADSeq locus discovery combined with amplicon sequencing of targeted loci is an effective method for incorporating paralogs into population genetic analyses.     

Paralogs are revealed by proportion of heterozygotes and deviations in read ratios in genotyping‐by‐sequencing data from natural populations

Whole‐genome duplications have occurred in the recent ancestors of many plants, fish, and amphibians, resulting in a pervasiveness of paralogous loci and the potential for both disomic and tetrasomic inheritance in the same genome. Paralogs can be difficult to reliably genotype and are often excluded from genotyping‐by‐sequencing (GBS) analyses; however, removal requires paralogs to be identified which is difficult without a reference genome. We present a method for identifying paralogs in natural populations by combining two properties of duplicated loci: (i) the expected frequency of heterozygotes exceeds that for singleton loci, and (ii) within heterozygotes, observed read ratios for each allele in GBS data will deviate from the 1:1 expected for singleton (diploid) loci. These deviations are often not apparent within individuals, particularly when sequence coverage is low; but, we postulated that summing allele reads for each locus over all heterozygous individuals in a population would provide sufficient power to detect deviations at those loci. We identified paralogous loci in three species: Chinook salmon (Oncorhynchus tshawytscha) which retains regions with ongoing residual tetrasomy on eight chromosome arms following a recent whole‐genome duplication, mountain barberry (Berberis alpina) which has a large proportion of paralogs that arose through an unknown mechanism, and dusky parrotfish (Scarus niger) which has largely rediploidized following an ancient whole‐genome duplication. Importantly, this approach only requires the genotype and allele‐specific read counts for each individual, information which is readily obtained from most GBS analysis pipelines.     

SNP discovery in wild and domesticated populations of blue catfish,Ictalurus furcatus,using genotyping‐by‐sequencing and subsequent SNP validation

Blue catfish, Ictalurus furcatus, are valued in the United States as a trophy fishery for their capacity to reach large sizes, sometimes exceeding 45 kg. Additionally, blue catfish × channel catfish (I. punctatus) hybrid food fish production has recently increased the demand for blue catfish broodstock. However, there has been little study of the genetic impacts and interaction of farmed, introduced and stocked populations of blue catfish. We utilized genotyping‐by‐sequencing (GBS) to capture and genotype SNP markers on 190 individuals from five wild and domesticated populations (Mississippi River, Missouri, D&B, Rio Grande and Texas). Stringent filtering of SNP‐calling parameters resulted in 4275 SNP loci represented across all five populations. Population genetics and structure analyses revealed potential shared ancestry and admixture between populations. We utilized the Sequenom MassARRAY to validate two multiplex panels of SNPs selected from the GBS data. Selection criteria included SNPs shared between populations, SNPs specific to populations, number of reads per individual and number of individuals genotyped by GBS. Putative SNPs were validated in the discovery population and in two additional populations not used in the GBS analysis. A total of 64 SNPs were genotyped successfully in 191 individuals from nine populations. Our results should guide the development of highly informative, flexible genotyping multiplexes for blue catfish from the larger GBS SNP set as well as provide an example of a rapid, low‐cost approach to generate and genotype informative marker loci in aquatic species with minimal previous genetic information.     

Estimation of population allele frequencies from next‐generation sequencing data: pool‐versus individual‐based genotyping

Molecular markers produced by next‐generation sequencing (NGS) technologies are revolutionizing genetic research. However, the costs of analysing large numbers of individual genomes remain prohibitive for most population genetics studies. Here, we present results based on mathematical derivations showing that, under many realistic experimental designs, NGS of DNA pools from diploid individuals allows to estimate the allele frequencies at single nucleotide polymorphisms (SNPs) with at least the same accuracy as individual‐based analyses, for considerably lower library construction and sequencing efforts. These findings remain true when taking into account the possibility of substantially unequal contributions of each individual to the final pool of sequence reads. We propose the intuitive notion of effective pool size to account for unequal pooling and derive a Bayesian hierarchical model to estimate this parameter directly from the data. We provide a user‐friendly application assessing the accuracy of allele frequency estimation from both pool‐ and individual‐based NGS population data under various sampling, sequencing depth and experimental error designs. We illustrate our findings with theoretical examples and real data sets corresponding to SNP loci obtained using restriction site–associated DNA (RAD) sequencing in pool‐ and individual‐based experiments carried out on the same population of the pine processionary moth (Thaumetopoea pityocampa). NGS of DNA pools might not be optimal for all types of studies but provides a cost‐effective approach for estimating allele frequencies for very large numbers of SNPs. It thus allows comparison of genome‐wide patterns of genetic variation for large numbers of individuals in multiple populations.     

Genetic sex assignment in wild populations using genotyping‐by‐sequencing data: A statistical threshold approach

Establishing the sex of individuals in wild systems can be challenging and often requires genetic testing. Genotyping‐by‐sequencing (GBS) and other reduced‐representation DNA sequencing (RRS) protocols (e.g., RADseq, ddRAD) have enabled the analysis of genetic data on an unprecedented scale. Here, we present a novel approach for the discovery and statistical validation of sex‐specific loci in GBS data sets. We used GBS to genotype 166 New Zealand fur seals (NZFS, Arctocephalus forsteri) of known sex. We retained monomorphic loci as potential sex‐specific markers in the locus discovery phase. We then used (i) a sex‐specific locus threshold (SSLT) to identify significantly male‐specific loci within our data set; and (ii) a significant sex‐assignment threshold (SSAT) to confidently assign sex in silico the presence or absence of significantly male‐specific loci to individuals in our data set treated as unknowns (98.9% accuracy for females; 95.8% for males, estimated via cross‐validation). Furthermore, we assigned sex to 86 individuals of true unknown sex using our SSAT and assessed the effect of SSLT adjustments on these assignments. From 90 verified sex‐specific loci, we developed a panel of three sex‐specific PCR primers that we used to ascertain sex independently of our GBS data, which we show amplify reliably in at least two other pinniped species. Using monomorphic loci normally discarded from large SNP data sets is an effective way to identify robust sex‐linked markers for nonmodel species. Our novel pipeline can be used to identify and statistically validate monomorphic and polymorphic sex‐specific markers across a range of species and RRS data sets.     

Genotype‐by‐sequencing of the plant‐pathogenic fungi Pyrenophora teres and Sphaerulina musiva utilizing Ion Torrent sequence technology

Genetic and genomics tools to characterize host–pathogen interactions are disproportionately directed to the host because of the focus on resistance. However, understanding the genetics of pathogen virulence is equally important and has been limited by the high cost of de novo genotyping of species with limited marker data. Non‐resource‐prohibitive methods that overcome the limitation of genotyping are now available through genotype‐by‐sequencing (GBS). The use of a two‐enzyme restriction‐associated DNA (RAD)‐GBS method adapted for Ion Torrent sequencing technology provided robust and reproducible high‐density genotyping of several fungal species. A total of 5783 and 2373 unique loci, ‘sequence tags’, containing 16 441 and 9992 single nucleotide polymorphisms (SNPs) were identified and characterized from natural populations of Pyrenophora teres f. maculata and Sphaerulina musiva, respectively. The data generated from the P. teres f. maculata natural population were used in association mapping analysis to map the mating‐type gene to high resolution. To further validate the methodology, a biparental population of P. teres f. teres, previously used to develop a genetic map utilizing simple sequence repeat (SSR) and amplified fragment length polymorphism (AFLP) markers, was re‐analysed using the SNP markers generated from this protocol. A robust genetic map containing 1393 SNPs on 997 sequence tags spread across 15 linkage groups with anchored reference markers was generated from the P. teres f. teres biparental population. The robust high‐density markers generated using this protocol will allow positional cloning in biparental fungal populations, association mapping of natural fungal populations and population genetics studies.     

Genotyping‐free parentage assignment using RAD‐seq reads

Parentage assignment is defined as the identification of the true parents of one focal offspring among a list of candidates and has been commonly used in zoological, ecological, and agricultural studies. Although likelihood‐based parentage assignment is the preferred method in most cases, it requires genotyping a predefined set of DNA markers and providing their population allele frequencies. In the present study, we proposed an alternative method of parentage assignment that does not depend on genotype data and prior information of allele frequencies. Our method employs the restriction site‐associated DNA sequencing (RAD‐seq) reads for clustering into the overlapped RAD loci among the compared individuals, following which the likelihood ratio of parentage assignment could be directly calculated using two parameters—the genome heterozygosity and error rate of sequencing reads. This method was validated on one simulated and two real data sets with the accurate assignment of true parents to focal offspring. However, our method could not provide a statistical confidence to conclude that the first ranked candidate is a true parent.     

A resource of genome‐wide single‐nucleotide polymorphisms generated by RAD tag sequencing in the critically endangered European eel

Reduced representation genome sequencing such as restriction‐site‐associated DNA (RAD) sequencing is finding increased use to identify and genotype large numbers of single‐nucleotide polymorphisms (SNPs) in model and nonmodel species. We generated a unique resource of novel SNP markers for the European eel using the RAD sequencing approach that was simultaneously identified and scored in a genome‐wide scan of 30 individuals. Whereas genomic resources are increasingly becoming available for this species, including the recent release of a draft genome, no genome‐wide set of SNP markers was available until now. The generated SNPs were widely distributed across the eel genome, aligning to 4779 different contigs and 19 703 different scaffolds. Significant variation was identified, with an average nucleotide diversity of 0.00529 across individuals. Results varied widely across the genome, ranging from 0.00048 to 0.00737 per locus. Based on the average nucleotide diversity across all loci, long‐term effective population size was estimated to range between 132 000 and 1 320 000, which is much higher than previous estimates based on microsatellite loci. The generated SNP resource consisting of 82 425 loci and 376 918 associated SNPs provides a valuable tool for future population genetics and genomics studies and allows for targeting specific genes and particularly interesting regions of the eel genome.     

Detection of individual ploidy levels with genotyping‐by‐sequencing (GBS) analysis

Ploidy levels sometimes vary among individuals or populations, particularly in plants. When such variation exists, accurate determination of cytotype can inform studies of ecology or trait variation and is required for population genetic analyses. Here, we propose and evaluate a statistical approach for distinguishing low‐level ploidy variants (e.g. diploids, triploids and tetraploids) based on genotyping‐by‐sequencing (GBS) data. The method infers cytotypes based on observed heterozygosity and the ratio of DNA sequences containing different alleles at thousands of heterozygous SNPs (i.e. allelic ratios). Whereas the method does not require prior information on ploidy, a reference set of samples with known ploidy can be included in the analysis if it is available. We explore the power and limitations of this method using simulated data sets and GBS data from natural populations of aspen (Populus tremuloides) known to include both diploid and triploid individuals. The proposed method was able to reliably discriminate among diploids, triploids and tetraploids in simulated data sets, and this was true for different levels of genetic diversity, inbreeding and population structure. Power and accuracy were minimally affected by low coverage (i.e. 2×), but did sometimes suffer when simulated mixtures of diploids, autotetraploids and allotetraploids were analysed. Cytotype assignments based on the proposed method closely matched those from previous microsatellite and flow cytometry data when applied to GBS data from aspen. An R package (gbs2ploidy) implementing the proposed method is available from CRAN.     

Helping decision making for reliable and cost‐effective 2b‐RAD sequencing and genotyping analyses in non‐model species

High‐throughput sequencing has revolutionized population and conservation genetics. RAD sequencing methods, such as 2b‐RAD, can be used on species lacking a reference genome. However, transferring protocols across taxa can potentially lead to poor results. We tested two different IIB enzymes (AlfI and CspCI) on two species with different genome sizes (the loggerhead turtle Caretta caretta and the sharpsnout seabream Diplodus puntazzo) to build a set of guidelines to improve 2b‐RAD protocols on non‐model organisms while optimising costs. Good results were obtained even with degraded samples, showing the value of 2b‐RAD in studies with poor DNA quality. However, library quality was found to be a critical parameter on the number of reads and loci obtained for genotyping. Resampling analyses with different number of reads per individual showed a trade‐off between number of loci and number of reads per sample. The resulting accumulation curves can be used as a tool to calculate the number of sequences per individual needed to reach a mean depth ≥20 reads to acquire good genotyping results. Finally, we demonstrated that selective‐base ligation does not affect genomic differentiation between individuals, indicating that this technique can be used in species with large genome sizes to adjust the number of loci to the study scope, to reduce sequencing costs and to maintain suitable sequencing depth for a reliable genotyping without compromising the results. Here, we provide a set of guidelines to improve 2b‐RAD protocols on non‐model organisms with different genome sizes, helping decision‐making for a reliable and cost‐effective genotyping.     

Population structure of Atlantic mackerel inferred from RAD‐seq‐derived SNP markers: effects of sequence clustering parameters and hierarchical SNP selection

Restriction‐site‐associated DNA sequencing (RAD‐seq) and related methods are revolutionizing the field of population genomics in nonmodel organisms as they allow generating an unprecedented number of single nucleotide polymorphisms (SNPs) even when no genomic information is available. Yet, RAD‐seq data analyses rely on assumptions on nature and number of nucleotide variants present in a single locus, the choice of which may lead to an under‐ or overestimated number of SNPs and/or to incorrectly called genotypes. Using the Atlantic mackerel (Scomber scombrus L.) and a close relative, the Atlantic chub mackerel (Scomber colias), as case study, here we explore the sensitivity of population structure inferences to two crucial aspects in RAD‐seq data analysis: the maximum number of mismatches allowed to merge reads into a locus and the relatedness of the individuals used for genotype calling and SNP selection. Our study resolves the population structure of the Atlantic mackerel, but, most importantly, provides insights into the effects of alternative RAD‐seq data analysis strategies on population structure inferences that are directly applicable to other species.     

Small‐ and large‐scale heterogeneity in genetic variation across the collard flycatcher genome: implications for estimating genetic diversity in nonmodel organisms

Population genetic studies in nonmodel organisms are often hampered by a lack of reference genomes that are essential for whole‐genome resequencing. In the light of this, genotyping methods have been developed to effectively eliminate the need for a reference genome, such as genotyping by sequencing or restriction site‐associated DNA sequencing (RAD‐seq). However, what remains relatively poorly studied is how accurately these methods capture both average and variation in genetic diversity across an organism's genome. In this issue of Molecular Ecology Resources, Dutoit et al. (2016) use whole‐genome resequencing data from the collard flycatcher to assess what factors drive heterogeneity in nucleotide diversity across the genome. Using these data, they then simulate how well different sequencing designs, including RAD sequencing, could capture most of the variation in genetic diversity. They conclude that for evolutionary and conservation‐related studies focused on the estimating genomic diversity, researchers should emphasize the number of loci analysed over the number of individuals sequenced.     

Double‐digest RAD sequencing using Ion Proton semiconductor platform (ddRADseq‐ion) with nonmodel organisms

Research in evolutionary biology involving nonmodel organisms is rapidly shifting from using traditional molecular markers such as mtDNA and microsatellites to higher throughput SNP genotyping methodologies to address questions in population genetics, phylogenetics and genetic mapping. Restriction site associated DNA sequencing (RAD sequencing or RADseq) has become an established method for SNP genotyping on Illumina sequencing platforms. Here, we developed a protocol and adapters for double‐digest RAD sequencing for Ion Torrent (Life Technologies; Ion Proton, Ion PGM) semiconductor sequencing. We sequenced thirteen genomic libraries of three different nonmodel vertebrate species on Ion Proton with PI chips: Arctic charr Salvelinus alpinus, European whitefish Coregonus lavaretus and common lizard Zootoca vivipara. This resulted in ~962 million single‐end reads overall and a mean of ~74 million reads per library. We filtered the genomic data using Stacks, a bioinformatic tool to process RAD sequencing data. On average, we obtained ~11 000 polymorphic loci per library of 6–30 individuals. We validate our new method by technical and biological replication, by reconstructing phylogenetic relationships, and using a hybrid genetic cross to track genomic variants. Finally, we discuss the differences between using the different sequencing platforms in the context of RAD sequencing, assessing possible advantages and disadvantages. We show that our protocol can be used for Ion semiconductor sequencing platforms for the rapid and cost‐effective generation of variable and reproducible genetic markers.     

Urine — a source for noninvasive genetic monitoring in wildlife

Noninvasive samples are of increasing importance to study wild populations. In this study, we investigate the applicability of urine samples as the sole source of DNA for routine noninvasive genetic monitoring of wildlife using wolves (Canis lupus) as an example. Within the scope of a long‐term wolf population survey, we collected during winter snow tracking in Bieszczady Mountains, Poland 41 urine samples considered as utilizable for genetic analyses. DNA concentration was determined by quantitative real‐time polymerase chain reaction (qPCR) and six microsatellite loci were genotyped in threefold repeated genotyping experiments to assess the reliability of genetic analyses of urine. DNA concentration of 33 urine samples was successfully quantified and of 14 samples, we obtained congruent results for all analysed loci and all repeated genotyping experiments. The gender of urine samples was identified with a Y‐chromosome‐linked marker. Considering the high discovery rate of urine in conjunction with its genotype reliability, our study confirms that urine is a valuable source in noninvasive genetic monitoring. Additionally, preselection of samples via qPCR proved to be a powerful tool contributing to a beneficial cost‐value ratio of genetic analyses by minimizing genotyping errors.     

Restriction site‐associated DNA sequencing,genotyping error estimation and de novo assembly optimization for population genetic inference

Restriction site‐associated DNA sequencing (RADseq) provides researchers with the ability to record genetic polymorphism across thousands of loci for nonmodel organisms, potentially revolutionizing the field of molecular ecology. However, as with other genotyping methods, RADseq is prone to a number of sources of error that may have consequential effects for population genetic inferences, and these have received only limited attention in terms of the estimation and reporting of genotyping error rates. Here we use individual sample replicates, under the expectation of identical genotypes, to quantify genotyping error in the absence of a reference genome. We then use sample replicates to (i) optimize de novo assembly parameters within the program Stacks, by minimizing error and maximizing the retrieval of informative loci; and (ii) quantify error rates for loci, alleles and single‐nucleotide polymorphisms. As an empirical example, we use a double‐digest RAD data set of a nonmodel plant species, Berberis alpina, collected from high‐altitude mountains in Mexico.     

The genetic architecture of tristyly and its breakdown to self‐fertilization

The floral polymorphism tristyly involves three style morphs with a reciprocal arrangement of stigma and anther heights governed by two diallelic loci (S and M). Tristyly functions to promote cross‐pollination, but modifications to stamen position commonly cause transitions to selfing. Here, we integrate whole‐genome sequencing and genetic mapping to investigate the genetic architecture of the M locus and the genetic basis of independent transitions to selfing in tristylous Eichhornia paniculata. We crossed independently derived semi‐homostylous selfing variants of the long‐ and mid‐styled morph fixed for alternate alleles at the M locus (ssmm and ssMM, respectively), and backcrossed the F₁ to the parental ssmm genotype. We phenotyped and genotyped 462 backcross progeny using 1450 genotyping‐by‐sequencing (GBS) markers and performed composite interval mapping to identify quantitative trait loci (QTL) governing style‐length and anther‐height variation. A QTL associated with the primary style‐morph differences (style length and anther height) mapped to linkage group 5 and spanned ~13–27.5 Mbp of assembled sequence. Bulk segregant analysis identified 334 genes containing SNPs potentially linked to the M locus. The stamen modifications characterizing each selfing variant were governed by loci on different linkage groups. Our results provide an important step towards identifying the M locus and demonstrate that transitions to selfing have originated by independent sets of mating‐system modifier genes unlinked to the M locus, a pattern inconsistent with a recombinational origin of selfing variants at a putative supergene.     

RAD‐sequencing for estimating genomic relatedness matrix‐based heritability in the wild: A case study in roe deer

Estimating the evolutionary potential of quantitative traits and reliably predicting responses to selection in wild populations are important challenges in evolutionary biology. The genomic revolution has opened up opportunities for measuring relatedness among individuals with precision, enabling pedigree‐free estimation of trait heritabilities in wild populations. However, until now, most quantitative genetic studies based on a genomic relatedness matrix (GRM) have focused on long‐term monitored populations for which traditional pedigrees were also available, and have often had access to knowledge of genome sequence and variability. Here, we investigated the potential of RAD‐sequencing for estimating heritability in a free‐ranging roe deer (Capreolous capreolus) population for which no prior genomic resources were available. We propose a step‐by‐step analytical framework to optimize the quality and quantity of the genomic data and explore the impact of the single nucleotide polymorphism (SNP) calling and filtering processes on the GRM structure and GRM‐based heritability estimates. As expected, our results show that sequence coverage strongly affects the number of recovered loci, the genotyping error rate and the amount of missing data. Ultimately, this had little effect on heritability estimates and their standard errors, provided that the GRM was built from a minimum number of loci (above 7,000). Genomic relatedness matrix‐based heritability estimates thus appear robust to a moderate level of genotyping errors in the SNP data set. We also showed that quality filters, such as the removal of low‐frequency variants, affect the relatedness structure of the GRM, generating lower h² estimates. Our work illustrates the huge potential of RAD‐sequencing for estimating GRM‐based heritability in virtually any natural population.     

Attack of the PCR clones: Rates of clonality have little effect on RAD‐seq genotype calls

Interpretation of high‐throughput sequence data requires an understanding of how decisions made during bioinformatic data processing can influence results. One source of bias that is often cited is PCR clones (or PCR duplicates). PCR clones are common in restriction site‐associated sequencing (RAD‐seq) data sets, which are increasingly being used for molecular ecology. To determine the influence PCR clones and the bioinformatic handling of clones have on genotyping, we evaluate four RAD‐seq data sets. Data sets were compared before and after clones were removed to estimate the number of clones present in RAD‐seq data, quantify how often the presence of clones in a data set causes genotype calls to change compared to when clones were removed, investigate the mechanisms that lead to genotype call changes and test whether clones bias heterozygosity estimates. Our RAD‐seq data sets contained 30%–60% PCR clones, but 95% of RAD‐tags had five or fewer clones. Relatively few genotypes changed once clones were removed (5%–10%), and the vast majority of these changes (98%) were associated with genotypes switching from a called to no‐call state or vice versa. PCR clones had a larger influence on genotype calls in individuals with low read depth but appeared to influence genotype calls at all loci similarly. Removal of PCR clones reduced the number of called genotypes by 2% but had almost no influence on estimates of heterozygosity. As such, while steps should be taken to limit PCR clones during library preparation, PCR clones are likely not a substantial source of bias for most RAD‐seq studies.