Novel R tools for analysis of genome-wide population genetic data with emphasis on clonality.
Kamvar ZN, Brooks JC, Grünwald NJ. · 2015
Read Novel R tools for analysis of genome-wide population genetic data with emphasis on clonality. by Kamvar ZN, Brooks JC, Grünwald NJ. (2015) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/europmc-pmc4462096.txt.
Introduction
Zhian Kamvar, Jonah Brooks, and Niklaus Grünwald added tools for population genetics of clonal organisms in Frontiers in Genetics in 2015. Plant microbes, the fungi and bacteria that attack crops, often reproduce clonally, copying themselves without mixing genes. Their populations need special analysis, and cheap genome sequencing now produces far larger data sets than older tools were built for.
The extensions, added to their R software package poppr, handle genome-wide marker data. New functions let users set clone boundaries and inspect lineages. They can draw networks that allow tangled links and scan along the genome for signs of gene mixing. They can also attach uncertainty estimates to any distance measure. Analyses can run at any level of a population hierarchy. The tools help researchers track how plant pathogens evolve and adapt to hosts and pesticides.
Catalog record
| CREATOR | Kamvar ZN, Brooks JC, Grünwald NJ. |
|---|---|
| YEAR | 2015 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 42 KB |
| LICENSE | CC BY 4.0 |
| RIGHTS | CC BY |
| SOURCE | Europe PMC |
Browse the full library · About the Tower of Babel
Full text
The opening of the work, from the library's hosted text:
Novel R tools for analysis of genome-wide population genetic data with emphasis on clonality By Zhian N. Kamvar, Jonah C. Brooks, Niklaus J. Grünwald Published in: Frontiers in Genetics DOI: 10.3389/fgene.2015.00208 ABSTRACT To gain a detailed understanding of how plant microbes evolve and adapt to hosts, pesticides, and other factors, knowledge of the population dynamics and evolutionary history of populations is crucial. Plant pathogen populations are often clonal or partially clonal which requires different analytical tools. With the advent of high throughput sequencing technologies, obtaining genome-wide population genetic data has become easier than ever before. We previously contributed the R package poppr specifically addressing issues with analysis of clonal populations. In this paper we provide several significant extensions to poppr with a focus on large, genome-wide SNP data. Specifically, we provide several new functionalities including the new function mlg.filter to define clone boundaries allowing for inspection and definition of what is a clonal lineage, minimum spanning networks with reticulation, a sliding-window analysis of the index of association, modular bootstrapping of any genetic distance, and analyses across any level of hierarchies. INTRODUCTION To paraphrase Dobzhansky, nothing in the field of plant-microbe interactions makes sense except in the light of population genetics (Dobzhansky, 1973). Genetic forces such as selection and drift act on alleles in a population. Thus, a true understanding of how plant pathogens emerge, evolve and adapt to crops, fungicides, or other factors, can only be elucidated in the context of population level phenomena given the demographic history of populations (Milgroom et al., 1989; McDonald and Linde, 2002; Grünwald and Goss, 2011). The field of population genetics, in the era of whole genome resequencing, provides unprecedented power to describe the evolutionary history and population processes that drive coevolution between pathogens and hosts. This powerful field thus critically enables effective deployment of R genes, design of pathogen informed plant resistance breeding programs, and implementation of fungicide rotations that minimize emergence of resistance. Most computational tools for population genetics are based on concepts developed for sexual model organisms. Populations that reproduce clonally or are polyploid are thus difficult to characterize using classical population genetic tools because theoretical assumptions underlying the theory are violated. Yet, many plant pathogen populations are at least partially clonal if not completely clonal (Anderson and Kohn, 1995; Milgroom, 1996). Thus, development of tools for analysis of clonal or polyploid populations is needed. Genotyping by sequencing and whole genome resequencing provide the unprecedented ability to identify thousands of single nucleotide polymorphisms (SNPs) in populations (Luikart et al., 2003; Davey et al., 2011; Elshire et al., 2011). With traditional marker data (e.g., SSR, AFLP) a clone was typically defined as a unique multilocus genotype (MLG) (Falush et al., 2003; Taylor and Fisher, 2003; Grünwald and Hoheisel, 2006; Goss et al., 2009; Cooke et al., 2012). Availability of large SNP data sets provides new challenges for data analysis. These data are based on reduced representation libraries and high throughput sequencing with moderate sequencing depth which invariably results in substantial missing data, error in SNP calling due to sequencing error, lack of read depth or other sources of spurious allele calls (Mastretta-Yanes et al., 2015). It is thus not clear what a clone is in large SNP data sets and novel tools are required for definition of clone boundaries. The research community using the R statistical and computing language (R Core Team, 2015) has developed a plethora of new resources for population genetic analysis. R is particularly appealing because all code is open source and functions can be evaluated and modified by any user. Recently, we introduced the R package poppr specifically developed for analysis of clonal populations (Kamvar et al., 2014b). Poppr previously introduced several novel features including the ability to conduct a hierarchical analysis across unlimited hierarchies, test for linkage association, graph minimum spanning networks or provide bootstrap support for Bruvo's distance in resulting trees. Poppr has been rapidly adopted and applied to a range of studies including for example horizontal transmission in leukemia of clams (Metzger et al., 2015), study of the vector-mediated parent-to-offspring transmission in an avian malaria-like parasite (Chakarov et al., 2015), and characterization of the emergence of the invasive forest pathogen Hymenoscyphus pseudoalbidus (Gross et al., 2014). It has also been used to implement real-time, online R based tools for visualizing relationships among unknown MLGs in reference databases (http://phytophthora-id.org/) (Grünwald et al., 2011). Here, we introduce poppr 2.0, which provides a major update to poppr (Kamvar et al., 2014b) including novel tools for analysis of clonal populations specifically addressing large SNP data. Significant novel tools include functions for calculating clone boundaries and collapsing individuals into clonal groups based on a user-specified genetic distance threshold, sliding window analyses, genotype accumulation curves, reticulations in minimum spanning networks, and bootstrapping for any genetic distance. IMPLEMENTATIONS AND EXAMPLES Clonal identification As highlighted in previous work, clone correction is an important component of population genetic analysis of organisms that are known to reproduce asexually (Milgroom, 1996; Grünwald et al., 2003; Kamvar et al., 2014b). This method is a partial correction for bias that affects metrics that rely on allele frequencies assuming panmixia and was initially designed for data with only a handful of markers. With the advent of large-scale sequencing and reduced- representation libraries, it has become easier to sequence tens of thousands of markers from hundreds of individuals (Davey and Blaxter, 2010; Davey et al., 2011; Elshire et al., 2011). With this larger number of markers, the genetic resolution is much greater, but the chance of genotyping error is also greatly increased and missing data is frequent (Mastretta-Yanes et al., 2015). Taking this fact and occasional somatic mutations into account, it would be impossible to separate true clones from independent individuals by just comparing what MLGs are different. We introduce a new method for collapsing unique multilocus genotypes determined by naive string comparison into multilocus lineages utilizing any genetic distance given three different clustering algorithms: farthest neighbor, nearest neighbor, and Unweighted Pair Group Method with Arithmetic Mean (UPGMA, average neighbor) (Sokal, 1958). These clustering algorithms act on a distance matrix that is either provided by the user or generated via a function that will calculate a distance from genetic data such as bruvo.dist, which in particular applies to any level of ploidy (Bruvo et al., 2004). All algorithms have been implemented in C and utilize the OpenMP framework for optional parallel processing (Dagum and Menon, 1998). Default is the conservative farthest neighbor algorithm (Figure 1A), which will only cluster samples together if all samples in the cluster are at a distance less than the given threshold. By contrast, the nearest neighbor algorithm will have a chaining effect that will cluster samples akin to adding links on a chain where a sample can be included in a cluster if all of the samples have at least one connection below a given threshold (Figure 1C). The UPGMA, or average neighbor clustering algorithm is the one most familiar to biologists as it is often used to generate ultra-metric trees based on genetic distance (Figure 1B). This algorithm will cluster by creating a representative sample per cluster and joining clusters if these representative samples are closer than the given threshold. [Figure: Diagrammatic representation of