PoPoolation: a toolbox for population genetic analysis of next generation sequencing data from pooled individuals.
Kofler R, Orozco-terWengel P, De Maio N, Pandey RV, Nolte V, Futschik A, Kosiol C, Schlötterer C. · 2011
Read PoPoolation: a toolbox for population genetic analysis of next generation sequencing data from pooled individuals. by Kofler R, Orozco-terWengel P, De Maio N, Pandey RV, Nolte V, Futschik A, Kosiol C, Schlötterer C. (2011) 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-pmc3017084.txt.
Introduction
Kofler and colleagues introduced PoPoolation, a toolbox for reading population genetics from pooled DNA samples, in a 2011 article. Sequencing many individuals mixed in one pool is cheap, but standard formulas misread such data.
Their programs estimate classic diversity and selection statistics. They correct for pooling bias and sequencing errors while doing so. They can also measure divergence between species. Results display as sliding window plots across the genome. Tests of mapping choices, error rates, and coverage show where pooled estimates stay trustworthy, helping smaller labs run genome-wide population studies within modest budgets.
Catalog record
| CREATOR | Kofler R, Orozco-terWengel P, De Maio N, Pandey RV, Nolte V, Futschik A, Kosiol C, Schlötterer C. |
|---|---|
| YEAR | 2011 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 37 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:
PoPoolation: A Toolbox for Population Genetic Analysis of Next Generation Sequencing Data from Pooled Individuals By Robert Kofler, Pablo Orozco-terWengel, Nicola De Maio, Ram Vinay Pandey, Viola Nolte, Andreas Futschik, Carolin Kosiol, Christian Schlötterer, Manfred Kayser Published in: PLoS ONE DOI: 10.1371/journal.pone.0015925 ABSTRACT Recent statistical analyses suggest that sequencing of pooled samples provides a cost effective approach to determine genome-wide population genetic parameters. Here we introduce PoPoolation, a toolbox specifically designed for the population genetic analysis of sequence data from pooled individuals. PoPoolation calculates estimates of θ Watterson, θ π, and Tajima's D that account for the bias introduced by pooling and sequencing errors, as well as divergence between species. Results of genome-wide analyses can be graphically displayed in a sliding window plot. PoPoolation is written in Perl and R and it builds on commonly used data formats. Its source code can be downloaded from http://code.google.com/p/popoolation/. Furthermore, we evaluate the influence of mapping algorithms, sequencing errors, and read coverage on the accuracy of population genetic parameter estimates from pooled data. INTRODUCTION The recent advances in sequencing technology have changed our experimental approaches to biological questions. It has become possible to move from small scale, gene centric studies to genome-wide analyses and remain within the budget of individual research grants. Even population genetic analyses have become within the reach of moderate research budgets by sequencing pools of individuals [e.g.: 1], [2] The new sequencing technologies have also changed the time allocation within a research project as well as the training required. Classic population studies typically involved a considerable wet-lab component for data collection. The new sequencing technologies reduce wet-lab work to DNA extraction and library construction. The analysis of the massive amounts of data generated in the course of a single experiment not only requires more time, but also new skills. The challenges of Next Generation Sequencing data, namely a hitherto unprecedented number of extremely short sequence reads containing more sequencing errors than previous sequencing technologies, have lead to the development of many new software tools over the past few years. For many applications, such as SNP (single nucleotide polymorphism) discovery [e.g.: 3], [4], [5], RNA-Seq [e.g.: 6], [7], ChIP-Seq [e.g.: 8], [9], and de novo assembly [e.g.: 10], [11], users can choose among a variety of software tools either in the public domain or from commercial software suppliers. For population genetic analyses, software tools are targeted at the analysis of individual genome sequencing projects [e.g.: 12]. To our knowledge no software packages are publicly available for population genetic analysis of pooled sequence data. Here, we introduce PoPoolation, a software suite specifically tailored for the analysis of pooled samples for population genetic inference. Furthermore, we carefully evaluate how peculiarities of the Next Generation Sequencing data (such as sequencing errors, mapping to a reference genome and read coverage) affect population genetic inferences. RESULTS The analysis of short sequence reads from pooled DNA samples requires several steps, as indicated in Figure 1. The first step in processing the data is trimming of the reads. Table 1 shows how trimming parameters influence the average length and quality of the reads used for mapping. While only few reads are lost with a quality threshold of 10 or 20, almost 70% are lost when a quality of 30 is used. We found that a threshold of 20 with a minimum length of 40–50 bp reliably generates high quality data. After trimming the reads for low quality bases, reads are mapped against a reference genome with the Burrows-Wheeler Alignment Tool (bwa [15]). Using SAMtools [15] the aligned reads are converted into a pileup file. This pileup file is used by PoPoolation to perform population genetic analyses. [Figure: Outline of a population genetic analysis from pooled sequence data.Sequencer figure from http://www.illumina.com/] [Table: Trimming statistics of 14×106 reads.] Validation Figure 2 shows the polymorphism and divergence pattern along the 3R chromosome of D. melanogaster. Our analysis captures important features of variability in D. melanogaster: regions close to the centromere and telomere (located at the left and right ends of figure 2, respectively) show the well-described drop in variability. [Figure: Graphical output of polymorphism and divergence estimates using PoPoolation.Sliding window analysis of θ π of a Portuguese D. melanogaster population on chromosome 3R (black line). The red line shows divergence (dxy) between D. melanogaster and D. simulans using the same window size and step size as for θ π. Note that dxy is scaled by 1/10. Both lines are based on non-overlapping windows of 50 kb.] Another striking feature apparent in Figure 2 is the ragged pattern of polymorphism, which shows 2-fold differences in variation between some regions in close physical proximity. To validate that this pattern reflects heterogeneity in sequence variation rather than problems with our pooling approach, we compared the polymorphism pattern on 3R obtained from a Portuguese D. melanogaster population to the polymorphism data generated by Sanger re-sequencing of a D. melanogaster population from The Netherlands [21]. Using the targeted regions option of PoPoolation (see below) we found a high correlation between our variability estimates and the ones published by Hutter et al. (2007) for the Dutch population (θ Watterson ρ = 0.78, p-value<2.2×10−16 and θ π ρ = 0.82, p-value<2.2×10−16; Table S1). Nevertheless, we also noted that the average variability in the Portuguese population was higher than for the Dutch population (θ Watterson: 0.0084 vs. 0.0065, Wilcoxon sum rank test p-value: 2.118*10−6; θ π: 0.0075 vs. 0.0063, Wilcoxon sum rank test p-value: 0.004567). Three important sources of error could affect the population genetic analysis of pooled samples: sequencing errors, problems with mapping the reads to the reference and insufficient sequence coverage. In the following, we evaluate all three factors. Sequencing errors The typical error rate of unprocessed reads from an Illumina sequencer is about 1%. As sequencing errors inevitably affect the polymorphism estimates and Tajima's D, it is highly desirable to reduce the sequencing error. It has been proposed to condition on a minor allele count larger than one, resulting in a truncated allele frequency spectrum [22]. We performed computer simulations to evaluate whether this correction is sufficient. We simulated 400 chromosomes (100 kb each) with the ms software [20]. The simulated chromosomes were re-coded into DNA sequence data by using a D. melanogaster chromosome as template. Finally, we generated random reads from these chromosomes with a sequencing error of 0.1–1%. These reads were then fed into the analysis pipeline of PoPoolation and the variability estimators were calculated. Our simulations show that with low error rates of 0.1–0.2% a minor allele count of two is well suited for a coverage up to 100, while for higher coverage a minor allele count of three is needed. Nevertheless, for an error rate of 1%, even a minor allele count of three is insufficient. Alternatively, it is possible to reduce the sequencing error by incorporating adequate quality control measures such that even low frequency alleles could be reliably detected and quantified [23], [24]. We evaluated whether simple quality measures could lead to a sufficient reduction in the error rates of Illumina reads to make their use in pooling experiments feasible. We determined the influence of trimming on the error rate of 74 bp (base pairs) Illumina sequence reads using