Germline variant calling using the seven bridges graph toolkit (SBGT)

A method for robustly and accurately detecting rare DNA mutations in tumor samples is critical to cancer research. Because many clinical tissue repositories have only FFPE-degraded tumor samples, and no matched normal sample from healthy tissue available, being able to discriminate low frequency mutations from background noise in the absence of a matched normal sample is of particular importance to research. Current state of the art variant callers such as GATK and VarScan focus on germline variant calling (used for detecting inherited mutations following a Mendelian inheritance pattern) or, in the case of FreeBayes and MuTect, focus on tumor-normal joint variant calling (using the normal sample to help discriminate low frequency somatic mutations from back ground noise). We present Pisces, a tumor-only variant caller exclusively developed at Illumina for detecting low frequency mutations from next generation sequencing data. Pisces has been an integral part of the Illumina Truseq Amplicon workflow since 2012, and is available on BaseSpace and on the MiSeq sequencing platforms. Pisces has been available to the public on github, since 2015. (https://github.com/Illumina/Pisces) Since that time, the Pisces variant calling team have continued to develop Pisces, and have made available a suite of variant calling tools, including a ReadStitcher, Variant Phaser, and Variant Quality Recalibration tool, to be used along with the core variant caller, Pisces. Here, we describe the Pisces variant calling tools and core algorithms. We describe the common use cases for Pisces (not necessarily restricted to somatic variant calling). We also evaluate Pisces performance on somatic and germline datasets, both from the titration of well characterized samples, and from a corpus of 500 FFPE-treated clinical trial tumor samples, against other variant callers. Our results show that Pisces gives highly accurate results in a variety of contexts. We recommend Pisces for amplicon somatic and germline variant calling.

BackgroundWhole exome sequencing (WES) is a cost-effective method that identifies clinical variants but it demands accurate variant caller tools. Currently available tools have variable accuracy in predicting specific clinical variants. But it may be possible to find the best combination of aligner-variant caller tools for detecting accurate single nucleotide variants (SNVs) and small insertion and deletion (InDels) separately. Moreover, many important aspects of InDel detection are overlooked while comparing the performance of tools, particularly its base pair length.ResultsWe assessed the performance of variant calling pipelines using the combinations of four variant callers and five aligners on human NA12878 and simulated exome data. We used high confidence variant calls from Genome in a Bottle (GiaB) consortium for validation, and GRCh37 and GRCh38 as the human reference genome. Based on the performance metrics, both BWA and Novoalign aligners performed better with DeepVariant and SAMtools callers for detecting SNVs, and with DeepVariant and GATK for InDels. Furthermore, we obtained similar results on human NA24385 and NA24631 exome data from GiaB.ConclusionIn this study, DeepVariant with BWA and Novoalign performed best for detecting accurate SNVs and InDels. The accuracy of variant calling was improved by merging the top performing pipelines. The results of our study provide useful recommendations for analysis of WES data in clinical genomics.

BackgroundPacific Biosciences (PacBio) circular consensus sequencing (CCS), also known as high fidelity (HiFi) technology, has revolutionized modern genomics by producing long (10 + kb) and highly accurate reads. This is achieved by sequencing circularized DNA molecules multiple times and combining them into a consensus sequence. Currently, the accuracy and quality value estimation provided by HiFi technology are more than sufficient for applications such as genome assembly and germline variant calling. However, there are limitations in the accuracy of the estimated quality scores when it comes to somatic variant calling on single reads.ResultsTo address the challenge of inaccurate quality scores for somatic variant calling, we introduce TopoQual, a novel tool designed to enhance the accuracy of base quality predictions. TopoQual leverages techniques including partial order alignments (POA), topologically parallel bases, and deep learning algorithms to polish consensus sequences. Our results demonstrate that TopoQual corrects approximately 31.9% of errors in PacBio consensus sequences. Additionally, it validates base qualities up to q59, which corresponds to one error in 0.9 million bases. These improvements will significantly enhance the reliability of somatic variant calling using HiFi data.ConclusionTopoQual represents a significant advancement in genomics by improving the accuracy of base quality predictions for PacBio HiFi sequencing data. By correcting a substantial proportion of errors and achieving high base quality validation, TopoQual enables confident and accurate somatic variant calling. This tool not only addresses a critical limitation of current HiFi technology but also opens new possibilities for precise genomic analysis in various research and clinical applications.

MotivationThe increasing adoption of clinical whole-genome resequencing (WGS) demands for highly accurate and reproducible variant calling (VC) methods. The observed discordance between state-of-the-art VC pipelines, however, indicates that the current practice still suffers from non-negligible numbers of false positive and negative SNV and INDEL calls that were shown to be enriched among discordant calls but also in genomic regions with low sequence complexity.ResultsHere, we describe our method ReliableGenome (RG) for partitioning genomes into high and low concordance regions with respect to a set of surveyed VC pipelines. Our method combines call sets derived by multiple pipelines from arbitrary numbers of datasets and interpolates expected concordance for genomic regions without data. By applying RG to 219 deep human WGS datasets, we demonstrate that VC concordance depends predominantly on genomic context rather than the actual sequencing data which manifests in high recurrence of regions that can/cannot be reliably genotyped by a single method. This enables the application of pre-computed regions to other data created with comparable sequencing technology and software. RG outperforms comparable efforts in predicting VC concordance and false positive calls in low-concordance regions which underlines its usefulness for variant filtering, annotation and prioritization. RG allows focusing resource-intensive algorithms (e.g. consensus calling methods) on the smaller, discordant share of the genome (20–30%) which might result in increased overall accuracy at reasonable costs. Our method and analysis of discordant calls may further be useful for development, benchmarking and optimization of VC algorithms and for the relative comparison of call sets between different studies/pipelines.Availability and ImplementationRG was implemented in Java, source code and binaries are freely available for non-commercial use at https://github.com/popitsch/wtchg-rg/.Supplementary information Supplementary data are available at Bioinformatics online.

BackgroundThe application of long-read sequencing using the Oxford Nanopore Technologies (ONT) MinION sequencer is getting more diverse in the medical field. Having a high sequencing error of ONT and limited throughput from a single MinION flowcell, however, limits its applicability for accurate variant detection. Medical exome sequencing (MES) targets clinically significant exon regions, allowing rapid and comprehensive screening of pathogenic variants. By applying MES with MinION sequencing, the technology can achieve a more uniform capture of the target regions, shorter turnaround time, and lower sequencing cost per sample.MethodWe introduced a cost-effective optimized workflow, ECNano, comprising a wet-lab protocol and bioinformatics analysis, for accurate variant detection at 4800 clinically important genes and regions using a single MinION flowcell. The ECNano wet-lab protocol was optimized to perform long-read target enrichment and ONT library preparation to stably generate high-quality MES data with adequate coverage. The subsequent variant-calling workflow, Clair-ensemble, adopted a fast RNN-based variant caller, Clair, and was optimized for target enrichment data. To evaluate its performance and practicality, ECNano was tested on both reference DNA samples and patient samples.ResultsECNano achieved deep on-target depth of coverage (DoC) at average > 100× and > 98% uniformity using one MinION flowcell. For accurate ONT variant calling, the generated reads sufficiently covered 98.9% of pathogenic positions listed in ClinVar, with 98.96% having at least 30× DoC. ECNano obtained an average read length of 1000 bp. The long reads of ECNano also covered the adjacent splice sites well, with 98.5% of positions having ≥ 30× DoC. Clair-ensemble achieved > 99% recall and accuracy for SNV calling. The whole workflow from wet-lab protocol to variant detection was completed within three days.ConclusionWe presented ECNano, an out-of-the-box workflow comprising (1) a wet-lab protocol for ONT target enrichment sequencing and (2) a downstream variant detection workflow, Clair-ensemble. The workflow is cost-effective, with a short turnaround time for high accuracy variant calling in 4800 clinically significant genes and regions using a single MinION flowcell. The long-read exon captured data has potential for further development, promoting the application of long-read sequencing in personalized disease treatment and risk prediction.

CanvasDB is an infrastructure for management and analysis of genetic variants from massively parallel sequencing (MPS) projects. The system stores SNP and indel calls in a local database, designed to handle very large datasets, to allow for rapid analysis using simple commands in R. Functional annotations are included in the system, making it suitable for direct identification of disease-causing mutations in human exome- (WES) or whole-genome sequencing (WGS) projects. The system has a built-in filtering function implemented to simultaneously take into account variant calls from all individual samples. This enables advanced comparative analysis of variant distribution between groups of samples, including detection of candidate causative mutations within family structures and genome-wide association by sequencing. In most cases, these analyses are executed within just a matter of seconds, even when there are several hundreds of samples and millions of variants in the database. We demonstrate the scalability of canvasDB by importing the individual variant calls from all 1092 individuals present in the 1000 Genomes Project into the system, over 4.4 billion SNPs and indels in total. Our results show that canvasDB makes it possible to perform advanced analyses of large-scale WGS projects on a local server.Database URL:https://github.com/UppsalaGenomeCenter/CanvasDB

Variant detection protocols for clinical next-generation sequencing (NGS) need application-specific optimization. Our aim was to analyze the performance of single nucleotide variant (SNV) and copy number (CNV) detection programs on an NGS panel for a rare disease. Thirty genes were sequenced in 83 patients with hereditary spastic paraplegia. The variant calls obtained with LifeScope, GATK UnifiedGenotyper and GATK HaplotypeCaller were compared with Sanger sequencing. The calling efficiency was evaluated for 187 (56 unique) SNVs and indels. Five multiexon deletions detected by multiple ligation probe assay were assessed from the NGS panel data with ExomeDepth, panelcn.MOPS and CNVPanelizer software. There were 48/51 (94%) SNVs and 1/5 (20%) indels consistently detected by all the calling algorithms. Two SNVs were not detected by any of the callers because of a rare reference allele, and one SNV in a low coverage region was only detected by two algorithms. Regarding CNVs, ExomeDepth detected 5/5 multi-exon deletions, panelcn.MOPs 4/5 and only 3/5 deletions were accurately detected by CNVPanelizer. The calling efficiency of NGS algorithms for SNVs is influenced by variant type and coverage. NGS protocols need to account for the presence of rare variants in the reference sequence as well as for ambiguities in indel calling. CNV detection algorithms can be used to identify large deletions from NGS panel data for diagnostic applications; however, sensitivity depends on coverage, selection of the reference set and deletion size. We recommend the incorporation of several variant callers in the NGS pipeline to maximize variant detection efficiency.

Heterozygous variants in the glucocerebrosidase GBA1 gene are an increasingly recognized risk factor for Parkinson’s disease (PD). Due to the GBAP1 pseudogene, which shares 96% sequence homology with the GBA1 coding region, accurate variant calling by array-based or short-read sequencing methods remains a major challenge in understanding the genetic landscape of GBA1-associated PD. We analyzed 660 patients with PD, 100 patients with Parkinsonism and 808 healthy controls from the Luxembourg Parkinson’s study, sequenced using amplicon-based long-read DNA sequencing technology. We found that 12.1% (77/637) of PD patients carried GBA1 variants, with 10.5% (67/637) of them carrying known pathogenic variants (including severe, mild, risk variants). In comparison, 5% (34/675) of the healthy controls carried GBA1 variants, and among them, 4.3% (29/675) were identified as pathogenic variant carriers. We found four GBA1 variants in patients with atypical parkinsonism. Pathogenic GBA1 variants were 2.6-fold more frequently observed in PD patients compared to controls (OR = 2.6; CI = [1.6,4.1]). Three novel variants of unknown significance (VUS) were identified. Using a structure-based approach, we defined a potential risk prediction method for VUS. This study describes the full landscape of GBA1-related parkinsonism in Luxembourg, showing a high prevalence of GBA1 variants as the major genetic risk for PD. Although the long-read DNA sequencing technique used in our study may be limited in its effectiveness to detect potential structural variants, our approach provides an important advancement for highly accurate GBA1 variant calling, which is essential for providing access to emerging causative therapies for GBA1 carriers.

Although it is ubiquitous in genomics, the current human reference genome (GRCh38) is incomplete: It is missing large sections of heterochromatic sequence, and as a singular, linear reference genome, it does not represent the full spectrum of human genetic diversity. To characterize gaps in GRCh38 and human genetic diversity, we developed an algorithm for sequence location approximation using nuclear families (ASLAN) to identify the region of origin of reads that do not align to GRCh38. Using unmapped reads and variant calls from whole-genome sequences (WGSs), ASLAN uses a maximum likelihood model to identify the most likely region of the genome that a subsequence belongs to given the distribution of the subsequence in the unmapped reads and phasings of families. Validating ASLAN on synthetic data and on reads from the alternative haplotypes in the decoy genome, ASLAN localizes >90% of 100-bp sequences with >92% accuracy and ∼1 Mb of resolution. We then ran ASLAN on 100-mers from unmapped reads from WGS from more than 700 families, and compared ASLAN localizations to alignment of the 100-mers to the recently released T2T-CHM13 assembly. We found that many unmapped reads in GRCh38 originate from telomeres and centromeres that are gaps in GRCh38. ASLAN localizations are in high concordance with T2T-CHM13 alignments, except in the centromeres of the acrocentric chromosomes. Comparing ASLAN localizations and T2T-CHM13 alignments, we identified sequences missing from T2T-CHM13 or sequences with high divergence from their aligned region in T2T-CHM13, highlighting new hotspots for genetic diversity.

After almost 30 years of inertia in the field of sequencing, the emergence of a whole range of so-called "next-generation" sequencing technologies has revolutionized the way we approach genomic and genetic research. Sequencing all 3 gigabases of a human genome, once a costly task of 13 years of international efforts, can now be done within a matter of days with a coverage of 30x and more, and comes with a price tag that is affordable for a middle-sized lab. Among the different next-generation sequencing machines developed over the course of the last 6 to 8 years, four instruments from three different companies have established themselves on the market for human whole-genome sequencing: Illumina's HiSeq2000, Life Technologies' SOLiD 4 and 5500xl SOLiD, and Complete Genomics' technology.
\n
\nHowever, these next-generation sequencing platforms are still relatively new, and a comprehensive comparative assessment of their performance is lacking. For this purpose, the DNA of two tumor-normal pairs from medulloblastoma patients was sequenced individually to 30x coverage on each of the four instruments. The resulting data was analyzed with respect to its coverage distribution and biases over the genome, in particular GC bias, and regions without coverage as well as specific genomic regions were assessed. SNP calls on the different sequencing machines were compared, and the benefits of combining read information from different instruments were evaluated. Additionally, somatic mutations were analyzed.
\n
\nThe most striking result is the poor coverage of GC-rich regions by SOLiD 4 and 5500xl SOLiD, discouraging their use in particular for methylation experiments and exome sequencing. In contrast, Complete Genomics seems the least affected by GC content and shows the most comprehensive coverage of many genomic regions, except for short repeats. HiSeq2000 exhibits the most even genome-wide coverage distribution and the least sample-to-sample variation, while consistently achieving the highest sensitivity in SNP calling. A combination of read data from different technologies is shown to entail limited improvement in most cases, and is advisable only for very specific applications. Finally, the comparison of somatic variation confirms that calling somatic alterations is still a big challenge, which is due in particular to low allele frequency. In summary, this comparative study illustrates the assets and drawbacks of each individual machine and can be used as a guide to find the most suitable platform for a specific experimental goal.

Computational methods are essential to extract actionable information from raw sequencing data, and to thus fulfill the promise of next-generation sequencing technology. Unfortunately, computational tools developed to call variants from human sequencing data disagree on many of their predictions, and current methods to evaluate accuracy and computational performance are ad hoc and incomplete. Agreement on benchmarking variant calling methods would stimulate development of genomic processing tools and facilitate communication among researchers. We propose SMaSH, a benchmarking methodology for evaluating germline variant calling algorithms. We generate synthetic datasets, organize and interpret a wide range of existing benchmarking data for real genomes and propose a set of accuracy and computational performance metrics for evaluating variant calling methods on these benchmarking data. Moreover, we illustrate the utility of SMaSH to evaluate the performance of some leading single-nucleotide polymorphism, indel and structural variant calling algorithms. We provide free and open access online to the SMaSH tool kit, along with detailed documentation, at smash.cs.berkeley.edu

Somatic variant calling involves the identification of genomic alterations that occur in somatic cells, requiring deep coverage to enable high sensitivity for low-frequency variants. Characterizing somatic variants across the entire genome therefore benefits from novel cost-efficient sequencing platforms, such as UG100. Here, we present optimization of variant calling tools for short and structural variants on WGS and WES data from UG100. For calling short variants, we optimized DeepVariant (DV) for somatic calling using data from matched tumor-normal sample pairs, improving both variant calling accuracy and pipeline running time (up to 10-fold). We defined the task of somatic variant calling as deciding if the pileup image containing reads from the tumor and normal samples represents a true somatic variant (vs a germline variant or artifact). The challenge of robust variant calling using deep learning models is exacerbated in somatic calling, where sequencing depth and coverage variability are typically high. Our optimized DV overcomes these challenges by several data sampling strategies. First, allele-frequency preserving down-sampling reduces randomness of read sub-sampling in high coverage regions. Second, alternative allele prioritization samples alt-allele supporting reads first allowing to call variants at very high coverage loci without sacrificing sensitivity and computational efficiency. Finally, a Panel-of-Normals based on targeted WES data provides an additional improvement of precision for this assay type. We used these strategies to train two models, one for tumor characterization using WGS (T/N coverage: 40x-150x/40x-100x), and one for deep WES (T/N coverage: >500x/>120x). We called variants on simulated tumors using the WGS model. For VAF>10% the model showed SNV recall >98% and indel recall >95% with false-positive rate of 0.2/Mb. For VAF range of 5-10%, indel recall was 67% and SNV recall was 86%. To demonstrate the utility of our somatic variant calling, we applied the models to call somatic variants from well characterized cancer cell lines: COLO829, HCC1395 and HCC1143. Results showed F1>90% for variants with VAF>10%. The WES model was used to reliably call variants at VAF>5% on simulated tumors with average SNV recall of 99% with precision >99% and indel recall >86% with precision >94%. To analyze structural and copy-number variations, we optimized the assembly engine of GRIDSS to enable fast calling of structural variations and demonstrate that Control-FREEC can be used to call copy number variants. SV calling on COLO829/COLO829BL achieved sensitivity >95%. In conclusion, our research highlights the utility of UG100 within the field of oncology, demonstrating its capacity for comprehensive and precise somatic variant detection, both on WGS and WES data. Citation Format: Doron Shem-Tov, Maya Levy, Gil Hornung, Ilya Soifer, Hila Benjamin, Ariel Jaimovich, Adam Blattler, William Brandler, Robert Sugar, Isaac Kinde, Omer Barad, Doron Lipson. Advancements in somatic variant calling from UG100 whole genome and whole exome sequencing data [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2024; Part 1 (Regular Abstracts); 2024 Apr 5-10; San Diego, CA. Philadelphia (PA): AACR; Cancer Res 2024;84(6_Suppl):Abstract nr 4926.

BackgroundTo promote the clinical application of next-generation sequencing, it is important to obtain accurate and consistent variants of target genomic regions at low cost. Ion Proton, the latest updated semiconductor-based sequencing instrument from Life Technologies, is designed to provide investigators with an inexpensive platform for human whole exome sequencing that achieves a rapid turnaround time. However, few studies have comprehensively compared and evaluated the accuracy of variant calling between Ion Proton and Illumina sequencing platforms such as HiSeq 2000, which is the most popular sequencing platform for the human genome. The Ion Proton sequencer combined with the Ion TargetSeq™ Exome Enrichment Kit together make up TargetSeq-Proton, whereas SureSelect-Hiseq is based on the Agilent SureSelect Human All Exon v4 Kit and the HiSeq 2000 sequencer.ResultsHere, we sequenced exonic DNA from four human blood samples using both TargetSeq-Proton and SureSelect-HiSeq. We then called variants in the exonic regions that overlapped between the two exome capture kits (33.6 Mb). The rates of shared variant loci called by two sequencing platforms were from 68.0 to 75.3 % in four samples, whereas the concordance of co-detected variant loci reached 99 %. Sanger sequencing validation revealed that the validated rate of concordant single nucleotide polymorphisms (SNPs) (91.5 %) was higher than the SNPs specific to TargetSeq-Proton (60.0 %) or specific to SureSelect-HiSeq (88.3 %). With regard to 1-bp small insertions and deletions (InDels), the Sanger sequencing validated rates of concordant variants (100.0 %) and SureSelect-HiSeq-specific (89.6 %) were higher than those of TargetSeq-Proton-specific (15.8 %).ConclusionsIn the sequencing of exonic regions, a combination of using of two sequencing strategies (SureSelect-HiSeq and TargetSeq-Proton) increased the variant calling specificity for concordant variant loci and the sensitivity for variant loci called by any one platform. However, for the sequencing of platform-specific variants, the accuracy of variant calling by HiSeq 2000 was higher than that of Ion Proton, specifically for the InDel detection. Moreover, the variant calling software also influences the detection of SNPs and, specifically, InDels in Ion Proton exome sequencing.Electronic supplementary materialThe online version of this article (doi:10.1186/s12864-015-1796-6) contains supplementary material, which is available to authorized users.

Analysis of NGS (Next Generation Sequencing) data is a computationally demanding task requiring large amounts of CPU, memory, and disk space. There is also a requirement for high performance data storage systems, resilient to hardware failure, to be connected directly to the computing infrastructure (typically a multi-node cluster) to store large quantities of NGS data reliably. Traditional shared file systems such as NFS (Network File System) do not offer the performance, scalability or cache coherence required by modern NGS data analysis, so alternatives including GlusterFS, Ceph, and Lustre have been developed. However, there is a trade-off between data safety on replicated local storage and degradation of performance across distributed storage. Resilience to hardware failure is typically provided by RAID (Redundant Array of Independent Disks) and redundant storage nodes. Here we describe the evaluation of an alternative file system, RozoFS ( https://github.com/rozofs/rozofs ) for use with demanding NGS data analysis workloads. We used a synthetic data set (DREAM-TCGA data set 3) to run a complete tumor-normal analysis pipeline (“bcbio”, https://github.com/chapmanb/bcbio-nextgen ), including base quality recalibration, local indel realignment, somatic variant calling, and structural variants as a benchmark to compare RozoFS with a traditional shared file system (NFS) on two different HPC (High Performance Computing; Cloud4CaRE project) clusters. Our results show high reliability and good performance of RozoFS compared to NFS, in particular, during heavy I/O workloads. These findings indicate that the reliability and robustness of RozoFS's make it a good candidate for demanding NGS analysis workloads.

Next-generation sequencing (NGS) has revolutionized the field of rare disease diagnostics. Whole exome and whole genome sequencing are now routinely used for diagnostic purposes; however, the overall diagnosis rate remains lower than expected. In this work, we review current approaches used for calling and interpretation of germline genetic variants in the human genome, and discuss the most important challenges that persist in the bioinformatic analysis of NGS data in medical genetics. We describe and attempt to quantitatively assess the remaining problems, such as the quality of the reference genome sequence, reproducible coverage biases, or variant calling accuracy in complex regions of the genome. We also discuss the prospects of switching to the complete human genome assembly or the human pan-genome and important caveats associated with such a switch. We touch on arguably the hardest problem of NGS data analysis for medical genomics, namely, the annotation of genetic variants and their subsequent interpretation. We highlight the most challenging aspects of annotation and prioritization of both coding and non-coding variants. Finally, we demonstrate the persistent prevalence of pathogenic variants in the coding genome, and outline research directions that may enhance the efficiency of NGS-based disease diagnostics.