Indel Calling Research Articles

Ultra-rapid emergency genomic diagnosis of Donahue syndrome in a preterm infant within 17 hours.

Genetic diseases are a major cause of neonatal morbidity and mortality. The clinical differential diagnosis in severely ill neonates, especially in premature infants, is challenging. Next generation sequencing (NGS) diagnostics is a valuable tool, but the turnaround time is often too long to provide a diagnosis in the time needed for clinical guidance in newborn intensive care units (NICU). To minimize turnaround time, we developed an ultra-rapid whole genome sequencing pipeline and tested it in clinical practice. Our pilot case, was a preterm infant presenting with several crises of dehydration, hypoglycaemia and hyponatremia together with nephrocalcinosis and hypertrophic cardiomyopathy. Whole genome sequencing was performed using a paired-end 2x75bp protocol. Sequencing data were exported after 50 sequencing cycles for a first analysis. After run completion, the rapid-sequencing protocol, a second analysis of the 2 x 75 paired-end run was performed. Both analyses comprised read-mapping and SNP-/indel calling on an on-site Edico Genome DRAGEN server, followed by functional annotation and pathogenicity prediction using in-house scripts. After the first analysis within 17 h, the emergency ultra-rapid protocol identified two novel compound heterozygous variants in the insulin receptor gene (INSR), pathogenic variants in which cause Donohue Syndrome. The genetic diagnosis could be confirmed by detection of hyperinsulinism and patient care adjusted. Nonetheless, we decided to pursue RNA studies, proving the functional effect of the novel splice variant and reduced expression levels of INSR in patients skin fibroblasts.

SICaRiO: short indel call filtering with boosting.

Despite impressive improvement in the next-generation sequencing technology, reliable detection of indels is still a difficult endeavour. Recognition of true indels is of prime importance in many applications, such as personalized health care, disease genomics and population genetics. Recently, advanced machine learning techniques have been successfully applied to classification problems with large-scale data. In this paper, we present SICaRiO, a gradient boosting classifier for the reliable detection of true indels, trained with the gold-standard dataset from 'Genome in a Bottle' (GIAB) consortium. Our filtering scheme significantly improves the performance of each variant calling pipeline used in GIAB and beyond. SICaRiO uses genomic features that can be computed from publicly available resources, i.e. it does not require sequencing pipeline-specific information (e.g. read depth). This study also sheds lights on prior genomic contexts responsible for the erroneous calling of indels made by sequencing pipelines. We have compared prediction difficulty for three categories of indels over different sequencing pipelines. We have also ranked genomic features according to their predictivity in determining false positives.

Removing reference bias and improving indel calling in ancient DNA data analysis by mapping to a sequence variation graph

BackgroundDuring the last decade, the analysis of ancient DNA (aDNA) sequence has become a powerful tool for the study of past human populations. However, the degraded nature of aDNA means that aDNA molecules are short and frequently mutated by post-mortem chemical modifications. These features decrease read mapping accuracy and increase reference bias, in which reads containing non-reference alleles are less likely to be mapped than those containing reference alleles. Alternative approaches have been developed to replace the linear reference with a variation graph which includes known alternative variants at each genetic locus. Here, we evaluate the use of variation graph software vg to avoid reference bias for aDNA and compare with existing methods.ResultsWe use vg to align simulated and real aDNA samples to a variation graph containing 1000 Genome Project variants and compare with the same data aligned with bwa to the human linear reference genome. Using vg leads to a balanced allelic representation at polymorphic sites, effectively removing reference bias, and more sensitive variant detection in comparison with bwa, especially for insertions and deletions (indels). Alternative approaches that use relaxed bwa parameter settings or filter bwa alignments can also reduce bias but can have lower sensitivity than vg, particularly for indels.ConclusionsOur findings demonstrate that aligning aDNA sequences to variation graphs effectively mitigates the impact of reference bias when analyzing aDNA, while retaining mapping sensitivity and allowing detection of variation, in particular indel variation, that was previously missed.

Abstract 5463: Accuracy improvements in somatic whole-genome small-variant calling with the DRAGEN platform

Abstract Introduction: Next-generation whole-genome sequencing promises to enable dramatic expansion of precision oncology and personalized cancer care. To support this effort, it is critical to develop computational tools that can analyze sequence data accurately and with rapid turn-around-time. One particularly challenging problem is the calling of somatic variants in matched tumor and normal samples. We present the DRAGENTM 3.5 somatic pipeline that performs 110x/40x whole-genome end-to-end analysis in under two hours with accuracy superior to that of all other tools we have benchmarked, including Mutect2/GATK4 and Strelka2. In addition, DRAGEN 3.5 is robust against variations in coverage, sequencing platform, sample preparation chemistry, and tumor purity. It also tolerates tumor-in-normal contamination, thereby making the pipeline applicable to late-stage solid tumors or hematological cancers. Methods: DRAGEN 3.5 replaces the legacy genotyping model (originally developed as part of MuTect2) with that of Strelka2. Whereas the MuTect2 model performs separate analyses on the tumor and normal samples, the Strelka2 model performs a joint analysis, allowing (1) detection of systematic errors that affect both samples simultaneously and (2) modeling of tumor-in-normal contamination, where the amount of contamination in the normal sample depends on the allele frequency observed in the tumor sample at the locus in question. In addition, we improved the probabilistic model of systematic error by incorporating models of strand bias and mismapping. Results: We benchmarked DRAGEN against Mutect2 (included in GATK 4.1.2) and Strelka2 (version 2.9.9) on five public synthetic and real datasets with known truth sets. DRAGEN greatly outperformed the other methods on all five datasets, producing 14-67% and 22-91% fewer false SNV calls, and 35-86% and 48-89% fewer false indel calls than Strelka2 and Mutect2 respectively. DRAGEN also exhibits higher tolerance to tumor-in-normal (TiN) contamination than Strelka2 which is already equipped with a model tolerating TiN contamination. The average end-to-end workflow runtime of the DRAGEN somatic pipeline was 77 minutes, 75% and 830% faster than Strelka2 and Mutect2 taking DRAGEN alignments as input. Conclusion: The DRAGEN pipeline enables reliable whole-genome analysis that can be scaled to large numbers of samples, leading to better tumor characterization and improved interpretation. We anticipate that it will ultimately fuel progress in oncology, cancer research and precision medicine. The DRAGEN 3.5 somatic pipeline can be run either locally on a DRAGEN server or remotely in the cloud via https://basespace.illumina.com. Citation Format: Konrad Scheffler, Sangtae Kim, Varun Jain, Jeffrey Yuan, Westley Sherman, Taylor O'Connell, Eric Ojard, Lisa Murray, Rami Mehio, Severine Catreux. Accuracy improvements in somatic whole-genome small-variant calling with the DRAGEN platform [abstract]. In: Proceedings of the Annual Meeting of the American Association for Cancer Research 2020; 2020 Apr 27-28 and Jun 22-24. Philadelphia (PA): AACR; Cancer Res 2020;80(16 Suppl):Abstract nr 5463.

Benchmarking variant callers in next-generation and third-generation sequencing analysis.

DNA variants represent an important source of genetic variations among individuals. Next- generation sequencing (NGS) is the most popular technology for genome-wide variant calling. Third-generation sequencing (TGS) has also recently been used in genetic studies. Although many variant callers are available, no single caller can call both types of variants on NGS or TGS data with high sensitivity and specificity. In this study, we systematically evaluated 11 variant callers on 12 NGS and TGS datasets. For germline variant calling, we tested DNAseq and DNAscope modes from Sentieon, HaplotypeCaller mode from GATK and WGS mode from DeepVariant. All the four callers had comparable performance on NGS data and 30× coverage of WGS data was recommended. For germline variant calling on TGS data, we tested DNAseq mode from Sentieon, HaplotypeCaller mode from GATK and PACBIO mode from DeepVariant. All the three callers had similar performance in SNP calling, while DeepVariant outperformed the others in InDel calling. TGS detected more variants than NGS, particularly in complex and repetitive regions. For somatic variant calling on NGS, we tested TNscope and TNseq modes from Sentieon, MuTect2 mode from GATK, NeuSomatic, VarScan2, and Strelka2. TNscope and Mutect2 outperformed the other callers. A higher proportion of tumor sample purity (from 10 to 20%) significantly increased the recall value of calling. Finally, computational costs of the callers were compared and Sentieon required the least computational cost. These results suggest that careful selection of a tool and parameters is needed for accurate SNP or InDel calling under different scenarios.

TSLP and IL-7R Variants Are Associated with Persistent Atopic Dermatitis

TSLP and IL-7R Variants Are Associated with Persistent Atopic Dermatitis

Comparison of somatic variant detection algorithms using Ion Torrent targeted deep sequencing data

BackgroundThe application of next-generation sequencing in cancer has revealed the genomic landscape of many tumour types and is nowadays routinely used in research and clinical settings. Multiple algorithms have been developed to detect somatic variation from sequencing data using either paired tumour-blood or tumour-only samples. Most of these methods have been developed and evaluated for the identification of somatic variation using Illumina sequencing datasets of moderate coverage. However, a comprehensive evaluation of somatic variant detection algorithms on Ion Torrent targeted deep sequencing data has not been performed.MethodsWe have applied three somatic detection algorithms, Torrent Variant Caller, MuTect2 and VarScan2, on a large cohort of ovarian cancer patients comprising of 208 paired tumour-blood samples and 253 tumour-only samples sequenced deeply on Ion Torrent Proton platform across 330 amplicons. Subsequently, the concordance and performance of the three somatic variant callers were assessed.ResultsWe have observed low concordance across the algorithms with only 0.5% of SNV and 0.02% of INDEL calls in common across all three methods. The intersection of all methods showed better performance when assessed using correlation with known mutational signatures, overlap with COSMIC variation and by examining the variant characteristics. The Torrent Variant Caller also performed well with the advantage of not eliminating a high number of variants that could lead to high type II error.ConclusionsOur results suggest that caution should be taken when applying state-of-the-art somatic variant algorithms to Ion Torrent targeted deep sequencing data. Better quality control procedures and strategies that combine results from multiple methods should ensure that higher accuracy is achieved. This is essential to ensure that results from bioinformatics pipelines using Ion Torrent deep sequencing can be robustly applied in cancer research and in the clinic.

2027P - VHIO-300 and a thousand and one nights: A tale of precision medicine

2027P - VHIO-300 and a thousand and one nights: A tale of precision medicine

DNA Sequencing Technologies

The recent advances in DNA sequencing technology, from first-generation sequencing (FGS) to third-generation sequencing (TGS), have constantly transformed the genome research landscape. Its data throughput is unprecedented and severalfold as compared with past technologies. DNA sequencing technologies generate sequencing data that are big, sparse, and heterogeneous. This results in the rapid development of various data protocols and bioinformatics tools for handling sequencing data.In this review, a historical snapshot of DNA sequencing is taken with an emphasis on data manipulation and tools. The technological history of DNA sequencing is described and reviewed in thorough detail. To manipulate the sequencing data generated, different data protocols are introduced and reviewed. In particular, data compression methods are highlighted and discussed to provide readers a practical perspective in the real-world setting. A large variety of bioinformatics tools are also reviewed to help readers extract the most from their sequencing data in different aspects, such as sequencing quality control, genomic visualization, single-nucleotide variant calling, INDEL calling, structural variation calling, and integrative analysis. Toward the end of the article, we critically discuss the existing DNA sequencing technologies for their pitfalls and potential solutions.

A new approach (EDIZ) for big data variant prioritization

Whole exome sequencing (WES), workflow consists of the following steps: raw data quality assessment, pre-processing, alignment, post-processing, variant calling, annotation, and prioritization. WES of human samples was reported to detect approximately 20,000–30,000 SNV and indel calls on average. Therefore, it is very important to choose the best tool that suits the related study. In this study, we aimed to upgrade our previous in-house variant prioritization method to analyse WES data without using in silico methods. By this method, the annotated data have been decreased by means of 52.3 times. Therefore, we both established a successful WES workflow for increasing the diagnostic rate of patients with reducing the raw data. Recently, we are also building a web-based workflow to help the users from all over the world.

Abstract 5130: Single-cell mutational profiling of paired AML samples at diagnosis, remission and relapse: Implications for therapeutic resistance and MRD detection

Abstract Acute myeloid leukemia (AML) is a molecularly heterogeneous disorder of the bone marrow with poor long-term clinical outcomes due to a high risk of relapse after initial therapy. Relapse is thought to occur from minimal residual disease (MRD) consisting of leukemic blasts present below the limit of morphologic detection. Genomic sequencing studies have suggested complex subclonal architecture of AML, however bulk sequencing techniques may not fully resolve individual subclones present at diagnosis, remission and relapse. Here, we use a high-throughput single cell sequencing technique to delineate the subclonal structure of AML and identify clones present at diagnosis and at time of remission associated with disease relapse.Matched diagnosis, remission, and relapse samples were examined for 20 de novo AML cases including 15 relapsed and 5 non-relapsed controls. Mutational bulk sequencing was performed by NGS panel sequencing and exome sequencing was available in select cases. Single cell processing was performed using the Tapestri platform (Mission Bio, inc). Briefly, individual cells were isolated using a microfluidic approach, followed by barcoding and genomic DNA amplification for individual cancer cells confined to droplets. Barcodes were then used to reassemble the genetic profiles of cells from next generation sequencing data. We applied this approach to individual AML samples, for accurate single nucleotide variant (SNV) and indel calling for up to 300 multiplexed loci from up to 10,000 cells in a single run.Targeted single-cell sequencing was able to recapitulate all mutations identified by bulk sequencing. Additionally, single cell analysis allowed for resolution of subclonal architecture and tumor phylogenetic evolution beyond what was predicted from bulk sequencing alone. We identified rare subclonal populations associated with relapse at time of diagnosis and remission. Additionally, single-cell sequencing unambiguously resolved pre-leukemic/clonal hematopoiesis-associated clones, which persisted following treatment. Analysis of paired samples at diagnosis and relapse identified subclonal populations below detection of standard bulk sequencing methods. These included rare single-cell populations associated with treatment resistance including independent subclones with IDH1 and IDH2 mutations, co-occurring RAS and FLT3 mutations, as well as persistent low level clones associated with MRD positivity and disease relapse. Our results suggest a greater degree of subclonal heterogeneity in de novo AML samples than inferred from bulk sequencing methods alone and shows the utility of single-cell sequencing for treatment monitoring, MRD identification, and understanding of disease resistance mechanisms. Citation Format: Alexey Aleshin, Robert Durruthy-Durruthy, Ryan Corces, Michaela Liedtke, Dennis Eastburn, Ravindra Majeti. Single-cell mutational profiling of paired AML samples at diagnosis, remission and relapse: Implications for therapeutic resistance and MRD detection [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2019; 2019 Mar 29-Apr 3; Atlanta, GA. Philadelphia (PA): AACR; Cancer Res 2019;79(13 Suppl):Abstract nr 5130.

Targeted capture-based NGS is superior to multiplex PCR-based NGS for hereditary BRCA1 and BRCA2 gene analysis in FFPE tumor samples

BackgroundWith the introduction of Olaparib treatment for BRCA-deficient recurrent ovarian cancer, testing for somatic and/or germline mutations in BRCA1/2 genes in tumor tissues became essential for treatment decisions. In most cases only formalin-fixed paraffin-embedded (FFPE) samples, containing fragmented and chemically modified DNA of minor quality, are available. Thus, multiplex PCR-based sequencing is most commonly applied in routine molecular testing, which is predominantly focused on the identification of known hot spot mutations in oncogenes.MethodsWe compared the overall performance of an adjusted targeted capture-based enrichment protocol and a multiplex PCR-based approach for calling of pathogenic SNVs and InDels using DNA extracted from 13 FFPE tissue samples. We further applied both strategies to seven blood samples and five matched FFPE tumor tissues of patients with known germline exon-spanning deletions and gene-wide duplications in BRCA1/2 to evaluate CNV detection based solely on panel NGS data. Finally, we analyzed DNA from FFPE tissues of 11 index patients from families suspected of having hereditary breast and ovarian cancer, of whom no blood samples were available for testing, in order to identify underlying pathogenic germline BRCA1/2 mutations.ResultsThe multiplex PCR-based protocol produced inhomogeneous coverage among targets of each sample and between samples as well as sporadic amplicon drop out, leading to insufficiently or non-covered nucleotides, which subsequently hindered variant detection. This protocol further led to detection of PCR-artifacts that could easily have been misinterpreted as pathogenic mutations. No such limitations were observed by application of an adjusted targeted capture-based protocol, which allowed for CNV calling with 86% sensitivity and 100% specificity. All pathogenic CNVs were confirmed in the five matched FFPE tumor samples from patients carrying known pathogenic germline mutations and we additionally identified somatic loss of the second allele in BRCA1/2. Furthermore we detected pathogenic BRCA1/2 variants in four the eleven FFPE samples from patients of whom no blood was available for analysis.ConclusionsWe demonstrate that an adjusted targeted capture-based enrichment protocol is superior to commonly applied multiplex PCR-based protocols for reliable BRCA1/2 variant detection, including CNV-detection, using FFPE tumor samples.

Empirical evaluation of variant calling accuracy using ultra-deep whole-genome sequencing data

In the design of whole-genome sequencing (WGS) studies, sequencing depth is a crucial parameter to define variant calling accuracy and study cost, with no standard recommendations having been established. We empirically evaluated the variant calling accuracy of the WGS pipeline using ultra-deep WGS data (approximately 410×). We randomly sampled sequence reads and constructed a series of simulation WGS datasets with a variety of gradual depths (n = 54; from 0.05× to 410×). Next, we evaluated the genotype concordances of the WGS data with those in the SNP microarray data or the WGS data using all the sequence reads. In addition, we assessed the accuracy of HLA allele genotyping using the WGS data with multiple software tools (PHLAT, HLA-VBseq, HLA-HD, and SNP2HLA). The WGS data with higher depths showed higher concordance rates, and >13.7× depth achieved as high as >99% of concordance. Comparisons with the WGS data using all the sequence reads showed that SNVs achieved >95% of concordance at 17.6× depth, whereas indels showed only 60% concordance. For the accuracy of HLA allele genotyping using the WGS data, 13.7× depth showed sufficient accuracy while performance heterogeneity among the software tools was observed (the highest concordance of 96.9% was observed with HLA-HD). Improvement in HLA genotyping accuracy by further increasing the depths was limited. These results suggest a medium degree of the WGS depth setting (approximately 15×) to achieve both accurate SNV calling and cost-effectiveness, whereas relatively higher depths are required for accurate indel calling.

Comparison of three variant callers for human whole genome sequencing

Testing of patients with genetics-related disorders is in progress of shifting from single gene assays to gene panel sequencing, whole-exome sequencing (WES) and whole-genome sequencing (WGS). Since WGS is unquestionably becoming a new foundation for molecular analyses, we decided to compare three currently used tools for variant calling of human whole genome sequencing data. We tested DeepVariant, a new TensorFlow machine learning-based variant caller, and compared this tool to GATK 4.0 and SpeedSeq, using 30×, 15× and 10× WGS data of the well-known NA12878 DNA reference sample. According to our comparison, the performance on SNV calling was almost similar in 30× data, with all three variant callers reaching F-Scores (i.e. harmonic mean of recall and precision) equal to 0.98. In contrast, DeepVariant was more precise in indel calling than GATK and SpeedSeq, as demonstrated by F-Scores of 0.94, 0.90 and 0.84, respectively. We conclude that the DeepVariant tool has great potential and usefulness for analysis of WGS data in medical genetics.

ICR142 Benchmarker: evaluating, optimising and benchmarking variant calling performance using the ICR142 NGS validation series.

Evaluating, optimising and benchmarking of next generation sequencing (NGS) variant calling performance are essential requirements for clinical, commercial and academic NGS pipelines. Such assessments should be performed in a consistent, transparent and reproducible fashion, using independently, orthogonally generated data. Here we present ICR142 Benchmarker, a tool to generate outputs for assessing germline base substitution and indel calling performance using the ICR142 NGS validation series, a dataset of Illumina platform-based exome sequence data from 142 samples together with Sanger sequence data at 704 sites.ICR142 Benchmarker provides summary and detailed information on the sensitivity, specificity and false detection rates of variant callers. ICR142 Benchmarker also automatically generates a single page report highlighting key performance metrics and how performance compares to widely-used open-source tools. We used ICR142 Benchmarker with VCF files outputted by GATK, OpEx and DeepVariant to create a benchmark for variant calling performance. This evaluation revealed pipeline-specific differences and shared challenges in variant calling, for example in detecting indels in short repeating sequence motifs. We next used ICR142 Benchmarker to perform regression testing with DeepVariant versions 0.5.2 and 0.6.1. This showed that v0.6.1 improves variant calling performance, but there was evidence of minor changes in indel calling behaviour that may benefit from attention. The data also allowed us to evaluate filters to optimise DeepVariant calling, and we recommend using 30 as the QUAL threshold for base substitution calls when using DeepVariant v0.6.1. Finally, we used ICR142 Benchmarker with VCF files from two commercial variant calling providers to facilitate optimisation of their in-house pipelines and to provide transparent benchmarking of their performance. ICR142 Benchmarker consistently and transparently analyses variant calling performance based on the ICR142 NGS validation series, using the standard VCF input and outputting informative metrics to enable user understanding of pipeline performance. ICR142 Benchmarker is freely available at https://github.com/RahmanTeamDevelopment/ICR142_Benchmarker/releases.

ICR142 Benchmarker: evaluating, optimising and benchmarking variant calling performance using the ICR142 NGS validation series

Evaluating, optimising and benchmarking of next generation sequencing (NGS) variant calling performance are essential requirements for clinical, commercial and academic NGS pipelines. Such assessments should be performed in a consistent, transparent and reproducible fashion, using independently, orthogonally generated data. Here we present ICR142 Benchmarker, a tool to generate outputs for assessing germline base substitution and indel calling performance using the ICR142 NGS validation series, a dataset of Illumina platform-based exome sequence data from 142 samples together with Sanger sequence data at 704 sites. ICR142 Benchmarker provides summary and detailed information on the sensitivity, specificity and false detection rates of variant callers. ICR142 Benchmarker also automatically generates a single page report highlighting key performance metrics and how performance compares to widely-used open-source tools. We used ICR142 Benchmarker with VCF files outputted by GATK, OpEx and DeepVariant to create a benchmark for variant calling performance. This evaluation revealed pipeline-specific differences and shared challenges in variant calling, for example in detecting indels in short repeating sequence motifs. We next used ICR142 Benchmarker to perform regression testing with DeepVariant versions 0.5.2 and 0.6.1. This showed that v0.6.1 improves variant calling performance, but there was evidence of minor changes in indel calling behaviour that may benefit from attention. The data also allowed us to evaluate filters to optimise DeepVariant calling, and we recommend using 30 as the QUAL threshold for base substitution calls when using DeepVariant v0.6.1. Finally, we used ICR142 Benchmarker with VCF files from two commercial variant calling providers to facilitate optimisation of their in-house pipelines and to provide transparent benchmarking of their performance. ICR142 Benchmarker consistently and transparently analyses variant calling performance based on the ICR142 NGS validation series, using the standard VCF input and outputting informative metrics to enable user understanding of pipeline performance. ICR142 Benchmarker is freely available at https://github.com/RahmanTeamDevelopment/ICR142_Benchmarker/releases.

The ICR142 NGS validation series: a resource for orthogonal assessment of NGS analysis.

To provide a useful community resource for orthogonal assessment of NGS analysis software, we present the ICR142 NGS validation series. The dataset includes high-quality exome sequence data from 142 samples together with Sanger sequence data at 704 sites; 416 sites with variants and 288 sites at which variants were called by an NGS analysis tool, but no variant is present in the corresponding Sanger sequence. The dataset includes 293 indel variants and 247 negative indel sites, and thus the ICR142 validation dataset is of particular utility in evaluating indel calling performance. The FASTQ files and Sanger sequence results can be accessed in the European Genome-phenome Archive under the accession number EGAS00001001332.

ICR142 Benchmarker: evaluating, optimising and benchmarking variant calling using the ICR142 NGS validation series.

Evaluating, optimising and benchmarking of next generation sequencing (NGS) variant calling performance are essential requirements for clinical, commercial and academic NGS pipelines. Such assessments should be performed in a consistent, transparent and reproducible fashion, using independently, orthogonally generated data. Here we present ICR142 Benchmarker, a tool to generate outputs for assessing variant calling performance using the ICR142 NGS validation series, a dataset of exome sequence data from 142 samples together with Sanger sequence data at 704 sites. ICR142 Benchmarker provides summary and detailed information on the sensitivity, specificity and false detection rates of variant callers. ICR142 Benchmarker also automatically generates a single page report highlighting key performance metrics and how performance compares to widely-used open-source tools. We used ICR142 Benchmarker with VCF files outputted by GATK, OpEx and DeepVariant to create a benchmark for variant calling performance. This evaluation revealed pipeline-specific differences and shared challenges in variant calling, for example in detecting indels in short repeating sequence motifs. We next used ICR142 Benchmarker to perform regression testing with versions 0.5.2 and 0.6.1 of DeepVariant. This showed that v0.6.1 improves variant calling performance, but there was evidence of some minor changes in indel calling behaviour that may benefit from attention in future updates. The data also allowed us to evaluate filters to optimise DeepVariant calling, and we recommend using 30 as the QUAL threshold for base substitution calls when using DeepVariant v0.6.1. Finally, we used ICR142 Benchmarker with VCF files from two commercial variant calling providers to facilitate optimisation of their in-house pipelines and to provide transparent benchmarking of their performance. ICR142 Benchmarker consistently and transparently analyses variant calling performance based on the ICR142 NGS validation series, using the standard VCF input and outputting informative metrics to enable user understanding of pipeline performance. ICR142 Benchmarker is freely available at https://github.com/RahmanTeamDevelopment/ICR142_Benchmarker/releases.

Abstract LB-382: Identification of predisposing genes for small bowel adenocarcinoma by exome sequencing

Abstract Small bowel adenocarcinoma (SBA) is a rare but aggressive cancer type with limited treatment options. Known predisposing factors include Crohn's disease, celiac disease, and hereditary syndromes such as familial adenomatous polyposis (FAP), Lynch syndrome, and Peutz-Jeghers syndrome. Here, our aim was to further characterize genetic susceptibility to SBA in a large population-based cohort and simultaneously demonstrate the ability to utilize tumor-only data to cost-effectively but reliably call germline variants. Information on all SBAs diagnosed in Finland between the years 2003-2011 were collected utilizing the Finnish Cancer Registry that maintains a nation-wide database on all cancer cases diagnosed in Finland since 1953. From these we selected all SBAs 1) confirmed as small bowel primary tumor, 2) with available tumor material, and 3) tumor content of at least 50%. Additionally, all relevant medical records were available for all cases. Altogether 106 tumors representing all three parts of the small bowel were selected for exome sequencing. The variant calls were produced with GATK HaplotypeCaller. Germline calls were extracted from the data by filtering out somatic calls that were originally produced by e.g. filtering SNV and indel calls against whole-genome and exome samples of the GnomAD dataset (n=138,632). To focus on possible disease-causing variants, the remaining putative germline variants with allele frequency >0.001 in the whole GnomAD and population-specific Finnish GnomAD set (n=1,747) were excluded. The germline origin of the observed, most prominent variants are being verified by Sanger sequencing, whenever corresponding normal DNA is available. First, we considered variants that were truncating or predicted damaging in silico in the 106 known cancer predisposing genes according to COSMIC. Eight of the 106 genes harbored such variants in at least two patients. We detected the pathogenic germline variant in all patients known to have a hereditary cancer syndrome (MLH1/MSH6 in three Lynch syndrome and APC in two FAP cases). We also identified two patients with a BRCA2 germline variant, one truncating and the other one predicted damaging. BRCA2 might play a role in SBA, thus far germline BRCA2 variants have been observed at least in cancers of the ampullary region. Next, we widened the analysis for other candidate genes but preliminary results show no clear candidates that would be shared by several patients. We will also look more closely into genes belonging to the same signaling pathways as the known syndrome causing genes. Finally, no single gene mutated in all the patients with celiac disease (n=10) was observed nor were there clear differences in the germlines of patients with and without celiac disease. This population-based study on predisposing variants in SBAs provides new information on their molecular genetic background, possibly having an impact also on their treatment. Citation Format: Ulrika A. Hänninen, Riku Katainen, Tomas Tanskanen, Jiri Hamberg, Ari Ristimäki, Eero Pukkala, Minna Taipale, Jukka-Pekka Mecklin, Mervi Aavikko, Linda M. Forsström, Esa Pitkänen, Netta Mäkinen, Lauri A. Aaltonen. Identification of predisposing genes for small bowel adenocarcinoma by exome sequencing [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2018; 2018 Apr 14-18; Chicago, IL. Philadelphia (PA): AACR; Cancer Res 2018;78(13 Suppl):Abstract nr LB-382.

Abstract 5294: CIMerge: A machine learning approach for merging and genotyping complex indel calls from NGS data

Abstract Complex insertion and deletion (complex indel) is a rare category of genomic structural variations, which is formed by inserting one or multiple DNA fragments into the genomic location where a deletion occurs. A recent study conducts a systematic analysis from over 8,000 pan-cancer cases, which reports hundreds of complex indels in cancer-associated genes, some of which are considered potentially druggable. Several approaches are proposed to detect complex indels from the new-generation sequencing data (both 2nd and 3rd generation). However, as different data-mining algorithms vary the preferences on capturing data patterns, different approaches may report the conflicted complex indel calls. Here, we propose a machine learning approach to correct the conflicted calls from different approaches and further estimate the genotype of each call. The proposed approach, implemented as CIMerge, adopts a relevance vector machine framework. For each candidate call, CIMerge first extracts a set of features on the candidate region, which includes the read depth, variant allelic frequency, number of the splitting/unmapped reads, number of the discordant paired-end reads, aligned contig, etc. CIMerge also considers another set of features on data-mining algorithm(s) that reported the candidate call, which includes the parameter settings, etc. Both sets of features are trained by the relevance vector machine framework, which outputs the probability of each candidate of the conflicted call. As a byproduct, it outputs the genotype of the candidate call with highest likelihood. We tested CIMerge on multiple datasets generated by different simulation configurations and compared it to several state-of-the-art approaches. The experiment results demonstrate that CIMerge outperforms the existing approaches. The average success rate of recognition is approaching 90%, while Pindel and Gindel are reported as 62.53% and 65.202%, respectively. The software package CIMerge is freely available for academic uses at https://github.com/xjtu712-lab/CIMerge. Citation Format: Tian Zheng, Yang Li, Yu Geng, Zhongmeng Zhao, Xuanping Zhang, Jiayin Wang. CIMerge: A machine learning approach for merging and genotyping complex indel calls from NGS data [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2018; 2018 Apr 14-18; Chicago, IL. Philadelphia (PA): AACR; Cancer Res 2018;78(13 Suppl):Abstract nr 5294.