GINDEL: Accurate Genotype Calling of Insertions and Deletions from Low Coverage Population Sequence Reads

Insertions and deletions (indels) are important types of structural variations. Obtaining accurate genotypes of indels may facilitate further genetic study. There are a few existing methods for calling indel genotypes from sequence reads. However, none of these tools can accurately call indel genotypes for indels of all lengths, especially for low coverage sequence data. In this paper, we present GINDEL, an approach for calling genotypes of both insertions and deletions from sequence reads. GINDEL uses a machine learning approach which combines multiple features extracted from next generation sequencing data. We test our approach on both simulated and real data and compare with existing tools, including Genome STRiP, Pindel and Clever-sv. Results show that GINDEL works well for deletions larger than 50 bp on both high and low coverage data. Also, GINDEL performs well for insertion genotyping on both simulated and real data. For comparison, Genome STRiP performs less well for shorter deletions (50–200 bp) on both simulated and real sequence data from the 1000 Genomes Project. Clever-sv performs well for intermediate deletions (200–1500 bp) but is less accurate when coverage is low. Pindel only works well for high coverage data, but does not perform well at low coverage. To summarize, we show that GINDEL not only can call genotypes of insertions and deletions (both short and long) for high and low coverage population sequence data, but also is more accurate and efficient than other approaches. The program GINDEL can be downloaded at: http://sourceforge.net/p/gindel     

One Size Doesn't Fit All - RefEditor: Building Personalized Diploid Reference Genome to Improve Read Mapping and Genotype Calling in Next Generation Sequencing Studies

With rapid decline of the sequencing cost, researchers today rush to embrace whole genome sequencing (WGS), or whole exome sequencing (WES) approach as the next powerful tool for relating genetic variants to human diseases and phenotypes. A fundamental step in analyzing WGS and WES data is mapping short sequencing reads back to the reference genome. This is an important issue because incorrectly mapped reads affect the downstream variant discovery, genotype calling and association analysis. Although many read mapping algorithms have been developed, the majority of them uses the universal reference genome and do not take sequence variants into consideration. Given that genetic variants are ubiquitous, it is highly desirable if they can be factored into the read mapping procedure. In this work, we developed a novel strategy that utilizes genotypes obtained a priori to customize the universal haploid reference genome into a personalized diploid reference genome. The new strategy is implemented in a program named RefEditor. When applying RefEditor to real data, we achieved encouraging improvements in read mapping, variant discovery and genotype calling. Compared to standard approaches, RefEditor can significantly increase genotype calling consistency (from 43% to 61% at 4X coverage; from 82% to 92% at 20X coverage) and reduce Mendelian inconsistency across various sequencing depths. Because many WGS and WES studies are conducted on cohorts that have been genotyped using array-based genotyping platforms previously or concurrently, we believe the proposed strategy will be of high value in practice, which can also be applied to the scenario where multiple NGS experiments are conducted on the same cohort. The RefEditor sources are available at https://github.com/superyuan/refeditor.

This is a PLOS Computational Biology Software Article.

     

XomAnnotate: Analysis of Heterogeneous and Complex Exome- A Step towards Translational Medicine

In translational cancer medicine, implicated pathways and the relevant master genes are of focus. Exome''s specificity, processing-time, and cost advantage makes it a compelling tool for this purpose. However, analysis of exome lacks reliable combinatory analysis tools and techniques. In this paper we present XomAnnotate – a meta- and functional-analysis software for exome. We compared UnifiedGenotyper, Freebayes, Delly, and Lumpy algorithms that were designed for whole-genome and combined their strengths in XomAnnotate for exome data through meta-analysis to identify comprehensive mutation profile (SNPs/SNVs, short inserts/deletes, and SVs) of patients. The mutation profile is annotated followed by functional analysis through pathway enrichment and network analysis to identify most critical genes and pathways implicated in the disease genesis. The efficacy of the software is verified through MDS and clustering and tested with available 11 familial non-BRCA1/BRCA2 breast cancer exome data. The results showed that the most significantly affected pathways across all samples are cell communication and antigen processing and presentation. ESCO1, HYAL1, RAF1 and PRKCA emerged as the key genes. Network analysis further showed the purine and propanotate metabolism pathways along with RAF1 and PRKCA genes to be master regulators in these patients. Therefore, XomAnnotate is able to use exome data to identify entire mutation landscape, pathways, and the master genes accurately with wide concordance from earlier microarray and whole-genome studies -- making it a suitable biomedical software for using exome in next-generation translational medicine.

Availability

http://www.iomics.in/research/XomAnnotate     

SavvyCNV: Genome-wide CNV calling from off-target reads

Identifying copy number variants (CNVs) can provide diagnoses to patients and provide important biological insights into human health and disease. Current exome and targeted sequencing approaches cannot detect clinically and biologically-relevant CNVs outside their target area. We present SavvyCNV, a tool which uses off-target read data from exome and targeted sequencing data to call germline CNVs genome-wide. Up to 70% of sequencing reads from exome and targeted sequencing fall outside the targeted regions. We have developed a new tool, SavvyCNV, to exploit this ‘free data’ to call CNVs across the genome. We benchmarked SavvyCNV against five state-of-the-art CNV callers using truth sets generated from genome sequencing data and Multiplex Ligation-dependent Probe Amplification assays. SavvyCNV called CNVs with high precision and recall, outperforming the five other tools at calling CNVs genome-wide, using off-target or on-target reads from targeted panel and exome sequencing. We then applied SavvyCNV to clinical samples sequenced using a targeted panel and were able to call previously undetected clinically-relevant CNVs, highlighting the utility of this tool within the diagnostic setting. SavvyCNV outperforms existing tools for calling CNVs from off-target reads. It can call CNVs genome-wide from targeted panel and exome data, increasing the utility and diagnostic yield of these tests. SavvyCNV is freely available at https://github.com/rdemolgen/SavvySuite.     

Comparison of solution-based exome capture methods for next generation sequencing

Background

Techniques enabling targeted re-sequencing of the protein coding sequences of the human genome on next generation sequencing instruments are of great interest. We conducted a systematic comparison of the solution-based exome capture kits provided by Agilent and Roche NimbleGen. A control DNA sample was captured with all four capture methods and prepared for Illumina GAII sequencing. Sequence data from additional samples prepared with the same protocols were also used in the comparison.

Results

We developed a bioinformatics pipeline for quality control, short read alignment, variant identification and annotation of the sequence data. In our analysis, a larger percentage of the high quality reads from the NimbleGen captures than from the Agilent captures aligned to the capture target regions. High GC content of the target sequence was associated with poor capture success in all exome enrichment methods. Comparison of mean allele balances for heterozygous variants indicated a tendency to have more reference bases than variant bases in the heterozygous variant positions within the target regions in all methods. There was virtually no difference in the genotype concordance compared to genotypes derived from SNP arrays. A minimum of 11× coverage was required to make a heterozygote genotype call with 99% accuracy when compared to common SNPs on genome-wide association arrays.

Conclusions

Libraries captured with NimbleGen kits aligned more accurately to the target regions. The updated NimbleGen kit most efficiently covered the exome with a minimum coverage of 20×, yet none of the kits captured all the Consensus Coding Sequence annotated exons.     

The eSNV-detect: a computational system to identify expressed single nucleotide variants from transcriptome sequencing data

Rapid development of next generation sequencing technology has enabled the identification of genomic alterations from short sequencing reads. There are a number of software pipelines available for calling single nucleotide variants from genomic DNA but, no comprehensive pipelines to identify, annotate and prioritize expressed SNVs (eSNVs) from non-directional paired-end RNA-Seq data. We have developed the eSNV-Detect, a novel computational system, which utilizes data from multiple aligners to call, even at low read depths, and rank variants from RNA-Seq. Multi-platform comparisons with the eSNV-Detect variant candidates were performed. The method was first applied to RNA-Seq from a lymphoblastoid cell-line, achieving 99.7% precision and 91.0% sensitivity in the expressed SNPs for the matching HumanOmni2.5 BeadChip data. Comparison of RNA-Seq eSNV candidates from 25 ER+ breast tumors from The Cancer Genome Atlas (TCGA) project with whole exome coding data showed 90.6–96.8% precision and 91.6–95.7% sensitivity. Contrasting single-cell mRNA-Seq variants with matching traditional multicellular RNA-Seq data for the MD-MB231 breast cancer cell-line delineated variant heterogeneity among the single-cells. Further, Sanger sequencing validation was performed for an ER+ breast tumor with paired normal adjacent tissue validating 29 out of 31 candidate eSNVs. The source code and user manuals of the eSNV-Detect pipeline for Sun Grid Engine and virtual machine are available at http://bioinformaticstools.mayo.edu/research/esnv-detect/.     

LoFreq: a sequence-quality aware,ultra-sensitive variant caller for uncovering cell-population heterogeneity from high-throughput sequencing datasets

The study of cell-population heterogeneity in a range of biological systems, from viruses to bacterial isolates to tumor samples, has been transformed by recent advances in sequencing throughput. While the high-coverage afforded can be used, in principle, to identify very rare variants in a population, existing ad hoc approaches frequently fail to distinguish true variants from sequencing errors. We report a method (LoFreq) that models sequencing run-specific error rates to accurately call variants occurring in <0.05% of a population. Using simulated and real datasets (viral, bacterial and human), we show that LoFreq has near-perfect specificity, with significantly improved sensitivity compared with existing methods and can efficiently analyze deep Illumina sequencing datasets without resorting to approximations or heuristics. We also present experimental validation for LoFreq on two different platforms (Fluidigm and Sequenom) and its application to call rare somatic variants from exome sequencing datasets for gastric cancer. Source code and executables for LoFreq are freely available at http://sourceforge.net/projects/lofreq/.     

Exome Sequencing of a Multigenerational Human Pedigree

Over the next few years, the efficient use of next-generation sequencing (NGS) in human genetics research will depend heavily upon the effective mechanisms for the selective enrichment of genomic regions of interest. Recently, comprehensive exome capture arrays have become available for targeting approximately 33 Mb or ∼180,000 coding exons across the human genome. Selective genomic enrichment of the human exome offers an attractive option for new experimental designs aiming to quickly identify potential disease-associated genetic variants, especially in family-based studies. We have evaluated a 2.1 M feature human exome capture array on eight individuals from a three-generation family pedigree. We were able to cover up to 98% of the targeted bases at a long-read sequence read depth of ≥3, 86% at a read depth of ≥10, and over 50% of all targets were covered with ≥20 reads. We identified up to 14,284 SNPs and small indels per individual exome, with up to 1,679 of these representing putative novel polymorphisms. Applying the conservative genotype calling approach HCDiff, the average rate of detection of a variant allele based on Illumina 1 M BeadChips genotypes was 95.2% at ≥10x sequence. Further, we propose an advantageous genotype calling strategy for low covered targets that empirically determines cut-off thresholds at a given coverage depth based on existing genotype data. Application of this method was able to detect >99% of SNPs covered ≥8x. Our results offer guidance for “real-world” applications in human genetics and provide further evidence that microarray-based exome capture is an efficient and reliable method to enrich for chromosomal regions of interest in next-generation sequencing experiments.     

Detection of Rare Genomic Variants from Pooled Sequencing Using SPLINTER

As DNA sequencing technology has markedly advanced in recent years², it has become increasingly evident that the amount of genetic variation between any two individuals is greater than previously thought³. In contrast, array-based genotyping has failed to identify a significant contribution of common sequence variants to the phenotypic variability of common disease^4,5. Taken together, these observations have led to the evolution of the Common Disease / Rare Variant hypothesis suggesting that the majority of the "missing heritability" in common and complex phenotypes is instead due to an individual''s personal profile of rare or private DNA variants^6-8. However, characterizing how rare variation impacts complex phenotypes requires the analysis of many affected individuals at many genomic loci, and is ideally compared to a similar survey in an unaffected cohort. Despite the sequencing power offered by today''s platforms, a population-based survey of many genomic loci and the subsequent computational analysis required remains prohibitive for many investigators.To address this need, we have developed a pooled sequencing approach^1,9 and a novel software package¹ for highly accurate rare variant detection from the resulting data. The ability to pool genomes from entire populations of affected individuals and survey the degree of genetic variation at multiple targeted regions in a single sequencing library provides excellent cost and time savings to traditional single-sample sequencing methodology. With a mean sequencing coverage per allele of 25-fold, our custom algorithm, SPLINTER, uses an internal variant calling control strategy to call insertions, deletions and substitutions up to four base pairs in length with high sensitivity and specificity from pools of up to 1 mutant allele in 500 individuals. Here we describe the method for preparing the pooled sequencing library followed by step-by-step instructions on how to use the SPLINTER package for pooled sequencing analysis (http://www.ibridgenetwork.org/wustl/splinter). We show a comparison between pooled sequencing of 947 individuals, all of whom also underwent genome-wide array, at over 20kb of sequencing per person. Concordance between genotyping of tagged and novel variants called in the pooled sample were excellent. This method can be easily scaled up to any number of genomic loci and any number of individuals. By incorporating the internal positive and negative amplicon controls at ratios that mimic the population under study, the algorithm can be calibrated for optimal performance. This strategy can also be modified for use with hybridization capture or individual-specific barcodes and can be applied to the sequencing of naturally heterogeneous samples, such as tumor DNA.     

Interpretation of custom designed Illumina genotype cluster plots for targeted association studies and next-generation sequence validation

Background

High-throughput custom designed genotyping arrays are a valuable resource for biologically focused research studies and increasingly for validation of variation predicted by next-generation sequencing (NGS) technologies. We investigate the Illumina GoldenGate chemistry using custom designed VeraCode and sentrix array matrix (SAM) assays for each of these applications, respectively. We highlight applications for interpretation of Illumina generated genotype cluster plots to maximise data inclusion and reduce genotyping errors.

Findings

We illustrate the dramatic effect of outliers in genotype calling and data interpretation, as well as suggest simple means to avoid genotyping errors. Furthermore we present this platform as a successful method for two-cluster rare or non-autosomal variant calling. The success of high-throughput technologies to accurately call rare variants will become an essential feature for future association studies. Finally, we highlight additional advantages of the Illumina GoldenGate chemistry in generating unusually segregated cluster plots that identify potential NGS generated sequencing error resulting from minimal coverage.

Conclusions

We demonstrate the importance of visually inspecting genotype cluster plots generated by the Illumina software and issue warnings regarding commonly accepted quality control parameters. In addition to suggesting applications to minimise data exclusion, we propose that the Illumina cluster plots may be helpful in identifying potential in-put sequence errors, particularly important for studies to validate NGS generated variation.

     

Performance of Genotype Imputation for Low Frequency and Rare Variants from the 1000 Genomes

Genotype imputation is now routinely applied in genome-wide association studies (GWAS) and meta-analyses. However, most of the imputations have been run using HapMap samples as reference, imputation of low frequency and rare variants (minor allele frequency (MAF) < 5%) are not systemically assessed. With the emergence of next-generation sequencing, large reference panels (such as the 1000 Genomes panel) are available to facilitate imputation of these variants. Therefore, in order to estimate the performance of low frequency and rare variants imputation, we imputed 153 individuals, each of whom had 3 different genotype array data including 317k, 610k and 1 million SNPs, to three different reference panels: the 1000 Genomes pilot March 2010 release (1KGpilot), the 1000 Genomes interim August 2010 release (1KGinterim), and the 1000 Genomes phase1 November 2010 and May 2011 release (1KGphase1) by using IMPUTE version 2. The differences between these three releases of the 1000 Genomes data are the sample size, ancestry diversity, number of variants and their frequency spectrum. We found that both reference panel and GWAS chip density affect the imputation of low frequency and rare variants. 1KGphase1 outperformed the other 2 panels, at higher concordance rate, higher proportion of well-imputed variants (info>0.4) and higher mean info score in each MAF bin. Similarly, 1M chip array outperformed 610K and 317K. However for very rare variants (MAF≤0.3%), only 0–1% of the variants were well imputed. We conclude that the imputation of low frequency and rare variants improves with larger reference panels and higher density of genome-wide genotyping arrays. Yet, despite a large reference panel size and dense genotyping density, very rare variants remain difficult to impute.     

Ultra-fast local-haplotype variant calling using paired-end DNA-sequencing data reveals somatic mosaicism in tumor and normal blood samples

Somatic mosaicism refers to the existence of somatic mutations in a fraction of somatic cells in a single biological sample. Its importance has mainly been discussed in theory although experimental work has started to emerge linking somatic mosaicism to disease diagnosis. Through novel statistical modeling of paired-end DNA-sequencing data using blood-derived DNA from healthy donors as well as DNA from tumor samples, we present an ultra-fast computational pipeline, LocHap that searches for multiple single nucleotide variants (SNVs) that are scaffolded by the same reads. We refer to scaffolded SNVs as local haplotypes (LH). When an LH exhibits more than two genotypes, we call it a local haplotype variant (LHV). The presence of LHVs is considered evidence of somatic mosaicism because a genetically homogeneous cell population will not harbor LHVs. Applying LocHap to whole-genome and whole-exome sequence data in DNA from normal blood and tumor samples, we find wide-spread LHVs across the genome. Importantly, we find more LHVs in tumor samples than in normal samples, and more in older adults than in younger ones. We confirm the existence of LHVs and somatic mosaicism by validation studies in normal blood samples. LocHap is publicly available at http://www.compgenome.org/lochap.     

Practical Issues in Imputation-Based Association Mapping

Imputation-based association methods provide a powerful framework for testing untyped variants for association with phenotypes and for combining results from multiple studies that use different genotyping platforms. Here, we consider several issues that arise when applying these methods in practice, including: (i) factors affecting imputation accuracy, including choice of reference panel; (ii) the effects of imputation accuracy on power to detect associations; (iii) the relative merits of Bayesian and frequentist approaches to testing imputed genotypes for association with phenotype; and (iv) how to quickly and accurately compute Bayes factors for testing imputed SNPs. We find that imputation-based methods can be robust to imputation accuracy and can improve power to detect associations, even when average imputation accuracy is poor. We explain how ranking SNPs for association by a standard likelihood ratio test gives the same results as a Bayesian procedure that uses an unnatural prior assumption—specifically, that difficult-to-impute SNPs tend to have larger effects—and assess the power gained from using a Bayesian approach that does not make this assumption. Within the Bayesian framework, we find that good approximations to a full analysis can be achieved by simply replacing unknown genotypes with a point estimate—their posterior mean. This approximation considerably reduces computational expense compared with published sampling-based approaches, and the methods we present are practical on a genome-wide scale with very modest computational resources (e.g., a single desktop computer). The approximation also facilitates combining information across studies, using only summary data for each SNP. Methods discussed here are implemented in the software package BIMBAM, which is available from http://stephenslab.uchicago.edu/software.html.     

Evaluating the Coverage and Potential of Imputing the Exome Microarray with Next-Generation Imputation Using the 1000 Genomes Project

Next-generation genotyping microarrays have been designed with insights from large-scale sequencing of exomes and whole genomes. The exome genotyping arrays promise to query the functional regions of the human genome at a fraction of the sequencing cost, thus allowing large number of samples to be genotyped. However, two pertinent questions exist: firstly, how representative is the content of the exome chip for populations not involved in the design of the chip; secondly, can the content of the exome chip be imputed with the reference data from the 1000 Genomes Project (1KGP). By deep whole-genome sequencing two Asian populations that are not part of the 1KGP, comprising 96 Southeast Asian Malays and 36 South Asian Indians for which the same samples have also been genotyped on both the Illumina 2.5 M and exome microarrays, we discovered the exome chip is a poor representation of exonic content in our two populations. However, up to 94.1% of the variants on the exome chip that are polymorphic in our populations can be confidently imputed with existing non-exome-centric microarrays using the 1KGP panel. The coverage further increases if there exists population-specific reference data from whole-genome sequencing. There is thus limited gain in using the exome chip for populations not involved in the microarray design. Instead, for the same cost of genotyping 2,000 samples on the exome chip, performing whole-genome sequencing of at least 35 samples in that population to complement the 1KGP may yield a higher coverage of the exonic content from imputation instead.     

Estimating autozygosity from high-throughput information: effects of SNP density and genotyping errors

Background

Runs of homozygosity are long, uninterrupted stretches of homozygous genotypes that enable reliable estimation of levels of inbreeding (i.e., autozygosity) based on high-throughput, chip-based single nucleotide polymorphism (SNP) genotypes. While the theoretical definition of runs of homozygosity is straightforward, their empirical identification depends on the type of SNP chip used to obtain the data and on a number of factors, including the number of heterozygous calls allowed to account for genotyping errors. We analyzed how SNP chip density and genotyping errors affect estimates of autozygosity based on runs of homozygosity in three cattle populations, using genotype data from an SNP chip with 777 972 SNPs and a 50 k chip.

Results

Data from the 50 k chip led to overestimation of the number of runs of homozygosity that are shorter than 4 Mb, since the analysis could not identify heterozygous SNPs that were present on the denser chip. Conversely, data from the denser chip led to underestimation of the number of runs of homozygosity that were longer than 8 Mb, unless the presence of a small number of heterozygous SNP genotypes was allowed within a run of homozygosity.

Conclusions

We have shown that SNP chip density and genotyping errors introduce patterns of bias in the estimation of autozygosity based on runs of homozygosity. SNP chips with 50 000 to 60 000 markers are frequently available for livestock species and their information leads to a conservative prediction of autozygosity from runs of homozygosity longer than 4 Mb. Not allowing heterozygous SNP genotypes to be present in a homozygosity run, as has been advocated for human populations, is not adequate for livestock populations because they have much higher levels of autozygosity and therefore longer runs of homozygosity. When allowing a small number of heterozygous calls, current software does not differentiate between situations where these calls are adjacent and therefore indicative of an actual break of the run versus those where they are scattered across the length of the homozygous segment. Simple graphical tests that are used in this paper are a current, yet tedious solution.     

Improved Ancestry Estimation for both Genotyping and Sequencing Data using Projection Procrustes Analysis and Genotype Imputation

Accurate estimation of individual ancestry is important in genetic association studies, especially when a large number of samples are collected from multiple sources. However, existing approaches developed for genome-wide SNP data do not work well with modest amounts of genetic data, such as in targeted sequencing or exome chip genotyping experiments. We propose a statistical framework to estimate individual ancestry in a principal component ancestry map generated by a reference set of individuals. This framework extends and improves upon our previous method for estimating ancestry using low-coverage sequence reads (LASER 1.0) to analyze either genotyping or sequencing data. In particular, we introduce a projection Procrustes analysis approach that uses high-dimensional principal components to estimate ancestry in a low-dimensional reference space. Using extensive simulations and empirical data examples, we show that our new method (LASER 2.0), combined with genotype imputation on the reference individuals, can substantially outperform LASER 1.0 in estimating fine-scale genetic ancestry. Specifically, LASER 2.0 can accurately estimate fine-scale ancestry within Europe using either exome chip genotypes or targeted sequencing data with off-target coverage as low as 0.05×. Under the framework of LASER 2.0, we can estimate individual ancestry in a shared reference space for samples assayed at different loci or by different techniques. Therefore, our ancestry estimation method will accelerate discovery in disease association studies not only by helping model ancestry within individual studies but also by facilitating combined analysis of genetic data from multiple sources.     

Inferring copy number and genotype in tumour exome data

Background

Using whole exome sequencing to predict aberrations in tumours is a cost effective alternative to whole genome sequencing, however is predominantly used for variant detection and infrequently utilised for detection of somatic copy number variation.

Results

We propose a new method to infer copy number and genotypes using whole exome data from paired tumour/normal samples. Our algorithm uses two Hidden Markov Models to predict copy number and genotypes and computationally resolves polyploidy/aneuploidy, normal cell contamination and signal baseline shift. Our method makes explicit detection on chromosome arm level events, which are commonly found in tumour samples. The methods are combined into a package named ADTEx (Aberration Detection in Tumour Exome). We applied our algorithm to a cohort of 17 in-house generated and 18 TCGA paired ovarian cancer/normal exomes and evaluated the performance by comparing against the copy number variations and genotypes predicted using Affymetrix SNP 6.0 data of the same samples. Further, we carried out a comparison study to show that ADTEx outperformed its competitors in terms of precision and F-measure.

Conclusions

Our proposed method, ADTEx, uses both depth of coverage ratios and B allele frequencies calculated from whole exome sequencing data, to predict copy number variations along with their genotypes. ADTEx is implemented as a user friendly software package using Python and R statistical language. Source code and sample data are freely available under GNU license (GPLv3) at http://adtex.sourceforge.net/.

Electronic supplementary material

The online version of this article (doi:10.1186/1471-2164-15-732) contains supplementary material, which is available to authorized users.