True Variants Research Articles

CAUSALdb: a database for disease/trait causal variants identified using summary statistics of genome-wide association studies.

Genome-wide association studies (GWASs) have revolutionized the field of complex trait genetics over the past decade, yet for most of the significant genotype-phenotype associations the true causal variants remain unknown. Identifying and interpreting how causal genetic variants confer disease susceptibility is still a big challenge. Herein we introduce a new database, CAUSALdb, to integrate the most comprehensive GWAS summary statistics to date and identify credible sets of potential causal variants using uniformly processed fine-mapping. The database has six major features: it (i) curates 3052 high-quality, fine-mappable GWAS summary statistics across five human super-populations and 2629 unique traits; (ii) estimates causal probabilities of all genetic variants in GWAS significant loci using three state-of-the-art fine-mapping tools; (iii) maps the reported traits to a powerful ontology MeSH, making it simple for users to browse studies on the trait tree; (iv) incorporates highly interactive Manhattan and LocusZoom-like plots to allow visualization of credible sets in a single web page more efficiently; (v) enables online comparison of causal relations on variant-, gene- and trait-levels among studies with different sample sizes or populations and (vi) offers comprehensive variant annotations by integrating massive base-wise and allele-specific functional annotations. CAUSALdb is freely available at http://mulinlab.org/causaldb.

Model-guided mechanism discovery and parameter selection for directed evolution.

Directed evolution is frequently applied to identify genetic variants with improvements in a single or multiple properties. When used to improve multiple properties simultaneously, a common strategy is to apply alternating rounds of selection criteria to enrich for variants with each desirable trait. In particular, counterselection, or selection against undesired traits rather than for desired ones, has been successfully employed in many studies. Although the sequence and stringency of alternating selective pressures for different traits are known to be highly consequential for the outcome of the screen, the effects of these parameters have not been systematically evaluated. We developed a method for producing a statistical modeling framework to elucidate these effects. The model uses single-cell fluorescence intensity distributions to estimate the proportions of phenotypic populations within a library and then predicts the changes in these proportions depending on specified positive selective or counterselective pressures. We validated the approach using recently described systems for metabolite-responsive bacterial transcription factors and yeast G-protein-coupled receptors. Finally, we applied the model to identify biological sources that exert undesirable selective pressure on libraries during sorting. Notably, these pressures produce substantial artifacts that, if unaddressed, can lead to failure of the screen. This method for model generation can be applied to FACS-based directed evolution experiments to create a quantitative framework that identifies subtle population effects. Such models can guide the choice of experimental design parameters to better enrich for true positive genetic variants and improve the chance of successful directed evolution.

Direct sequencing of RNA with MinION Nanopore: detecting mutations based on associations.

One of the key challenges in the field of genetics is the inference of haplotypes from next generation sequencing data. The MinION Oxford Nanopore sequencer allows sequencing long reads, with the potential of sequencing complete genes, and even complete genomes of viruses, in individual reads. However, MinION suffers from high error rates, rendering the detection of true variants difficult. Here, we propose a new statistical approach named AssociVar, which differentiates between true mutations and sequencing errors from direct RNA/DNA sequencing using MinION. Our strategy relies on the assumption that sequencing errors will be dispersed randomly along sequencing reads, and hence will not be associated with each other, whereas real mutations will display a non-random pattern of association with other mutations. We demonstrate our approach using direct RNA sequencing data from evolved populations of the MS2 bacteriophage, whose small genome makes it ideal for MinION sequencing. AssociVar inferred several mutations in the phage genome, which were corroborated using parallel Illumina sequencing. This allowed us to reconstruct full genome viral haplotypes constituting different strains that were present in the sample. Our approach is applicable to long read sequencing data from any organism for accurate detection of bona fide mutations and inter-strain polymorphisms.

Electronation-dependent structural change at the proton exit side of cytochrome c oxidase as revealed by site-directed fluorescence labeling.

Cytochrome c oxidase (CcO), the terminal enzyme of the respiratory chain of mitochondria and many aerobic prokaryotes that function as a redox-coupled proton pump, catalyzes the reduction of molecular oxygen to water. As part of the respiratory chain, CcO contributes to the proton motive force driving ATP synthesis. While many aspects of the enzyme's catalytic mechanisms have been established, a clear picture of the proton exit pathway(s) remains elusive. Here, we aim to gain insight into the molecular mechanisms of CcO through the development of a new homologous mutagenesis/expression system in Paracoccusdenitrificans, which allows mutagenesis of CcO subunits 1, 2, and 3. Our system provides true single thiol-reactive CcO variants in a three-subunit base variant with unique labeling sites for the covalent attachment of reporter groups sensitive to nanoenvironmental factors like protonation, polarity, and hydration. To this end, we exchanged six residues on both membrane sides of CcO for cysteines. We show redox-dependent wetting changes at the proton uptake channel and increased polarity at the proton exit side of CcO upon electronation. We suggest an electronation-dependent conformational change to play a role in proton exit from CcO.

Rare variants in non-coding regulatory regions of the genome that affect gene expression in systemic lupus erythematosus

Personalized medicine approaches are increasingly sought for diseases with a heritable component. Systemic lupus erythematosus (SLE) is the prototypic autoimmune disease resulting from loss of immunologic tolerance, but the genetic basis of SLE remains incompletely understood. Genome wide association studies (GWAS) identify regions associated with disease, based on common single nucleotide polymorphisms (SNPs) within them, but these SNPs may simply be markers in linkage disequilibrium with other, causative mutations. Here we use an hierarchical screening approach for prediction and testing of true functional variants within regions identified in GWAS; this involved bioinformatic identification of putative regulatory elements within close proximity to SLE SNPs, screening those regions for potentially causative mutations by high resolution melt analysis, and functional validation using reporter assays. Using this approach, we screened 15 SLE associated loci in 143 SLE patients, identifying 7 new variants including 5 SNPs and 2 insertions. Reporter assays revealed that the 5 SNPs were functional, altering enhancer activity. One novel variant was linked to the relatively well characterized rs9888739 SNP at the ITGAM locus, and may explain some of the SLE heritability at this site. Our study demonstrates that non-coding regulatory elements can contain private sequence variants affecting gene expression, which may explain part of the heritability of SLE.

Denoising of Aligned Genomic Data

Noise in genomic sequencing data is known to have effects on various stages of genomic data analysis pipelines. Variant identification is an important step of many of these pipelines, and is increasingly being used in clinical settings to aid medical practices. We propose a denoising method, dubbed SAMDUDE, which operates on aligned genomic data in order to improve variant calling performance. Denoising human data with SAMDUDE resulted in improved variant identification in both individual chromosome as well as whole genome sequencing (WGS) data sets. In the WGS data set, denoising led to identification of almost 2,000 additional true variants, and elimination of over 1,500 erroneously identified variants. In contrast, we found that denoising with other state-of-the-art denoisers significantly worsens variant calling performance. SAMDUDE is written in Python and is freely available at https://github.com/ihwang/SAMDUDE.

The complexity of screening PMS2 in DNA isolated from formalin-fixed paraffin-embedded material.

Germline variants in the DNA mismatch repair (MMR) gene PMS2 cause 1-14% of all Lynch Syndrome cancers. Correct variant analysis of PMS2 is complex due to the presence of multiple pseudogenes and the occurrence of gene conversion. The analysis complexity increases in highly fragmented DNA from formalin-fixed paraffin-embedded (FFPE) tissue. Here we describe a reliable approach to detect true PMS2 variants in fragmented DNA. A custom NGS panel designed for FFPE tissue was used targeting four MMR genes, POLE and POLD1. Amplicon design for PMS2 was based on the position of paralogous sequence variants (PSVs) that distinguish PMS2 from its pseudogenes. PMS2 variants in exons 1-11 can be correctly curated based on this information. For exons 12-15 this is less reliable as these undergo gene conversion. Using this method, we screened PMS2 variants in 125 MMR-deficient tumors. Of the 125 tumors tested, six were unexplained MMR-deficient tumors with solitary PMS2 protein expression loss. In these six tumors two unclassified variants (class 3) and five variants likely affecting function (class 4/5) were detected in PMS2. One microsatellite unstable tumor with positive staining for all MMR proteins was found to carry a frameshift PMS2 variant (class 5). No class 4 or class 5 PMS2 variants were detected in tumors with other patterns of MMR protein expression loss.

A novel computational OMICS and non-OMICS approach for identifying true pathogenic risk variants for Asian prostate cancer.

47 Background: Large-scale genome-wide association studies have established germline polygenic risk loci that underpin the susceptibility to prostate cancer (PCa). However, most trials conducted are in men of European ancestry with data missing for Asian male PCa. Here, we report on an in-house multidimensional bioinformatics pipeline that integrates OMICS and non-OMICS approaches in identifying true germline risk-variants for PCa in Asian men. Methods: We utilized a prospective cohort study of Asian men who were newly diagnosed with PCa. Whole exome sequencing (Illumina Hiseq, CA) of blood (100X) was performed. The OMICS-based approach entailed a stepwise screen for hallmarks of cancer-specific pathways. A genome-proteome network was then developed to filter for known pathogenic variants; this was followed by comparison against a large artificial database of aggregated germline variants (N = 95,000) with reported linkage to PCa susceptibility. Finally, mutations were filtered through a non-OMICS pipeline that entailed data synchronization with population-level statistics and clinical outcomes (recurrence and survival). Results: Preliminary analyses were based on 277 PCa cases; of which 50 were M1 cases. Screening using a non-combined unbiased approach yielded 36,157 germline variants. This contrast against our OMICS-based approach, which reduced the variant calls to 6,144 significantly associated mutations. Next, by focusing on pathway-specific genes related to hormonal regulation and known cancer hotspot mutations, we could further tighten our variant calls to 3,562 hormone-related variants (rs9269958 on HLA-DRB1) and 2,125 variants in known cancer genes, notably (rs8176320 on BRCA1/2, rs2555691 on LILRA2, rs8036934 on TP53BP1). Conclusions: Here, we show that application of an OMICS approach that combines pathway-driven analyses and an artificial dataset, along with population-level statistics and clinical relevance resulted in more robust annotation of germline variants that were associated with PCa.

Immune dysregulation underpins radioresistance in nasopharyngeal carcinoma (NPC).

52 Background: Radiotherapy (RT) is a primary modality in the treatment of NPC. However, 30% of patients present with disease recurrence following RT of this radiosensitive tumor. Here, we investigated the molecular and immune profiles associated with radioresistant (RR) NPC. Additionally, we investigated for aberrant molecular pathways in paired recurrences of patients to uncover new drivers underpinning radioresistance. Methods: We prospectively recruited a cohort of 100 NPC patients who completed definitive RT/chemoRT; including 30 cases who were recruited at recurrence. Whole exome sequencing (WES) at 200x was performed to identify low frequencies ( < 1%) of true somatic nucleotide variants (SNVs) and copy number alterations (CNAs). Transcriptomic profiles from RNAseq were interrogated using supervised and unsupervised statistical approaches to determine aberrant pathways that were significantly associated with RR. Results: Genomic instabilityas characterized by percentage genome alteration (PGA) was comparable in our cohort. Additionally, we did not observe any common or exclusive CNAs between RR- and nr-NPC cases. Based on a constellation of immune-related signatures, we observed an “immune-cold” profile that is associated with RR-NPC compared to nr-NPC controls, which is characterized by low expression of CD8+ T cell infiltration and interferon-γ response. Expectedly, pathways relating to angiogenesis, hypoxia and NOTCH signaling were upregulated in the RR-NPC cohort. Interestingly, we observed a reversal of the immune phenotype from “cold” to an enrichment of effector T cell infiltration in the paired recurrences. Conclusions: Here, we present a comprehensive mutational landscape of RR-NPC, which revealed the potential role of the immune environment in modulating RR. The longitudinal immune dysregulation of the tumor microenvironment between the de novo tumors and recurrences could be a driver or passenger event during the onset of recurrence.

P5723IEVA: Integration and Extraction of Variant Attributes in NGS analysis

Abstract Next Generation Sequencing (NGS) technology has taken a central role in the diagnosis of genetically-determined cardiovascular diseases. The differentiation of sequencing errors from real variants is a key-point of the genetic testing. In case of novel variants or variants of uncertain significance that may potentially impact on clinical decisions (e.g. ICD implantation in primary cardiomyopathies or preventive surgery in heritable aneurysmal diseases) it is crucial to exclude false positive (FP) and false negative (FN) sequence errors. To date, Sanger Sequencing is the gold standard tool used to confirm and validate NGS-identified variants. While FPs are excluded by the Sanger confirmation and the damage is for wasted costs and time, the FNs are non-resolvable errors because they are undetected and, obviously, not searched in Sanger confirmation, with the risk of missing genetic diagnoses. Purpose This project aimed at reducing NGS errors through the introduction of a bioinformatic solution in the bioinformatic analytic step of the genetic testing process. iEVA is a tool that enhances NGS-derived informative features to use them in a filtering process based on a Machine Learning algorithm (ML). It considers sequencing features (e.g. technical errors, duplicates of PCR) together with nucleotide sequence characteristics. Methods To demonstrate the effectiveness of iEVA in eliminating FP and FN errors from the NGS bioinformatic pipeline, we developed two ML-based filtering algorithms. The training dataset consisted of 7968 single nucleotide variants (SNV) and 306 Insertions and Deletions (InDels) validated by Sanger sequencing performed by expert molecular biologists. Variants derived from 800 sequences obtained with the Illumina Trusight Cardio panel containing 174 genes related to cardiovascular diseases. Two Random Forest classifiers were trained with the task of discriminating between sequencing error and real variant. The first one was trained using attributes derived from the most common variant caller (GATK v3.8), and the second one using iEVA results. To evaluate ML models, we used a 3-Fold cross-Validation and validated the results using an independent validation dataset consisting of 3415 SNV and 132 InDels. Results Using iEVA attributes, we obtained 1 FP (excluded by Sanger) and 3 FN (confirmed by Sanger) less than using common variant caller attributes. In the independent validation dataset, the iEVA-trained classification model identified 1 Sanger-confirmed variant that was missed by variant caller-trained model. Conclusions Variant filtering is crucial to exclude sequencing errors and to recognize true variants. Even a single filtering error may negatively impact on the patient when a genetic diagnosis is missed. To obtain a certain genetic diagnosis, a 0% error probability is needed. The introduction of iEVA in the pipeline is an easy, time- and cost-saving tool to reduce errors and to improve the precision of the genetic data. Acknowledgement/Funding Italian Ministry of Health Research Funding to the IRCCS Foundation University Hospital Policlinico San Matteo of Pavia

Benchmarking variant identification tools for plant diversity discovery

BackgroundThe ability to accurately and comprehensively identify genomic variations is critical for plant studies utilizing high-throughput sequencing. Most bioinformatics tools for processing next-generation sequencing data were originally developed and tested in human studies, raising questions as to their efficacy for plant research. A detailed evaluation of the entire variant calling pipeline, including alignment, variant calling, variant filtering, and imputation was performed on different programs using both simulated and real plant genomic datasets.ResultsA comparison of SOAP2, Bowtie2, and BWA-MEM found that BWA-MEM was consistently able to align the most reads with high accuracy, whereas Bowtie2 had the highest overall accuracy. Comparative results of GATK HaplotypCaller versus SAMtools mpileup indicated that the choice of variant caller affected precision and recall differentially depending on the levels of diversity, sequence coverage and genome complexity. A cross-reference experiment of S. lycopersicum and S. pennellii reference genomes revealed the inadequacy of single reference genome for variant discovery that includes distantly-related plant individuals. Machine-learning-based variant filtering strategy outperformed the traditional hard-cutoff strategy resulting in higher number of true positive variants and fewer false positive variants. A 2-step imputation method, which utilized a set of high-confidence SNPs as the reference panel, showed up to 60% higher accuracy than direct LD-based imputation.ConclusionsPrograms in the variant discovery pipeline have different performance on plant genomic dataset. Choice of the programs is subjected to the goal of the study and available resources. This study serves as an important guiding information for plant biologists utilizing next-generation sequencing data for diversity characterization and crop improvement.

FERMI: A Novel Method for Sensitive Detection of Rare Mutations in Somatic Tissue.

With growing interest in monitoring mutational processes in normal tissues, tumor heterogeneity, and cancer evolution under therapy, the ability to accurately and economically detect ultra-rare mutations is becoming increasingly important. However, this capability has often been compromised by significant sequencing, PCR and DNA preparation error rates. Here, we describe FERMI (Fast Extremely Rare Mutation Identification) - a novel method designed to eliminate the majority of these sequencing and library-preparation errors in order to significantly improve rare somatic mutation detection. This method leverages barcoded targeting probes to capture and sequence DNA of interest with single copy resolution. The variant calls from the barcoded sequencing data are then further filtered in a position-dependent fashion against an adaptive, context-aware null model in order to distinguish true variants. As a proof of principle, we employ FERMI to probe bone marrow biopsies from leukemia patients, and show that rare mutations and clonal evolution can be tracked throughout cancer treatment, including during historically intractable periods like minimum residual disease. Importantly, FERMI is able to accurately detect nascent clonal expansions within leukemias in a manner that may facilitate the early detection and characterization of cancer relapse.

Identification of somatic mutations in single cell DNA-seq using a spatial model of allelic imbalance

Recent advances in single cell technology have enabled dissection of cellular heterogeneity in great detail. However, analysis of single cell DNA sequencing data remains challenging due to bias and artifacts that arise during DNA extraction and whole-genome amplification, including allelic imbalance and dropout. Here, we present a framework for statistical estimation of allele-specific amplification imbalance at any given position in single cell whole-genome sequencing data by utilizing the allele frequencies of heterozygous single nucleotide polymorphisms in the neighborhood. The resulting allelic imbalance profile is critical for determining whether the variant allele fraction of an observed mutation is consistent with the expected fraction for a true variant. This method, implemented in SCAN-SNV (Single Cell ANalysis of SNVs), substantially improves the identification of somatic variants in single cells. Our allele balance framework is broadly applicable to genotype analysis of any variant type in any data that might exhibit allelic imbalance.

Long-read amplicon denoising.

Long-read next-generation amplicon sequencing shows promise for studying complete genes or genomes from complex and diverse populations. Current long-read sequencing technologies have challenging error profiles, hindering data processing and incorporation into downstream analyses. Here we consider the problem of how to reconstruct, free of sequencing error, the true sequence variants and their associated frequencies from PacBio reads. Called ‘amplicon denoising’, this problem has been extensively studied for short-read sequencing technologies, but current solutions do not always successfully generalize to long reads with high indel error rates. We introduce two methods: one that runs nearly instantly and is very accurate for medium length reads and high template coverage, and another, slower method that is more robust when reads are very long or coverage is lower. On two Mock Virus Community datasets with ground truth, each sequenced on a different PacBio instrument, and on a number of simulated datasets, we compare our two approaches to each other and to existing algorithms. We outperform all tested methods in accuracy, with competitive run times even for our slower method, successfully discriminating templates that differ by a just single nucleotide. Julia implementations of Fast Amplicon Denoising (FAD) and Robust Amplicon Denoising (RAD), and a webserver interface, are freely available.

Development and Application of Duplex Sequencing Strategy for Cell-Free DNA–Based Longitudinal Monitoring of Stage IV Colorectal Cancer

Potential applications of cell-free DNA (cfDNA)-based molecular profiling have used in patients with diverse malignant tumors. However, capturing all cfDNA that originates from tumor cells and identifying true variants present in this minute fraction remain challenges to the widespread application of cfDNA-based liquid biopsies in the clinical setting. In this study, we evaluate a systematic approach and identify key components of wet bench and bioinformatics strategies to address these challenges. We found that concentration of enrichment oligonucleotides, elements of the library preparation, and the structure of adaptors are critical for achieving high enrichment of target regions, retaining variant allele frequencies accurately throughout all involved steps of library preparation, and obtaining high variant coverage. We developed a dual molecular barcode-integrated error elimination strategy to remove sequencing artifacts and a background error correction strategy to distinguish true variants from abundant false-positive variants. We further describe a clinical application of this cfDNA-based duplex sequencing approach that can be used to monitor disease progression in patients with stage IV colorectal cancer. The findings also suggest that cfDNA-based molecular testing observations are highly concordant with observations obtained by traditional imaging methods. Overall, the findings presented in this study have potential implications for early detection of cancer, identification of minimal residualdisease, and evaluation of therapeutic responses in patients with cancer.

Abstract 1706: High-quality whole-genome sequencing and analysis of FFPE samples from multiplecancer types

Abstract Next-generation sequencing analysis of formalin-fixed, paraffin-embedded (FFPE) samples has the potential to lead to major advances in cancer treatment and prevention. However, whole-genome sequencing (WGS) of these samples has remained challenging due to the difficult process of isolating high-quality DNA and distinguishing true variant calls from artifacts. As evidence of these challenges, only four WGS studies of FFPE samples have been published to date. Previous work has attempted to overcome these challenges by using whole-exome sequencing or smaller exon panels at high coverage depths with publications reporting success from less than half to most samples. Thus, there has not been widespread adoption of using FFPE samples for next-generation sequencing. We developed a novel DNA extraction and library prep protocol for whole-genome sequencing of FFPE tissue samples. This method produces larger DNA fragments that can be successfully used for WGS. We previously presented results of a large esophageal cancer cohort where we sequenced over 1,000 samples in tumor-normal pairs with data equivalent to that generated from fresh-frozen tissues. We recently expanded the tumor types, ages, and sources and achieved similar results. We have assayed FFPE tissues derived from breast cancer, colorectal cancer, non-small cell lung cancer, prostate cancer, and melanoma. These samples were sourced from several different labs, which have used variations on the standard fixation protocol. Some of these FFPE samples were over 15 years old. Variant calling provides excellent results without excessive C>T mutations. The coverage uniformity was very good, and CNV calls were made with results comparable to published TCGA data. These samples have been analyzed for MSI and TMB, and the results are comparable to those using fresh tissues. Our method makes new types of clinical studies possible utilizing FFPE samples that have been generated as part of standard hospital procedures. This method also facilitates performing studies comparing primary and recurrent cancers easier to establish and run as recurrent cancer cases can be enrolled and stored FFPE primary samples used to establish a cancer baseline. Citation Format: Shannon T. Bailey, James Lund, Muhammad Ekram, Wanfeng Yu, Richard T. Williams, Hongye Sun, Jeffrey R. Gulcher. High-quality whole-genome sequencing and analysis of FFPE samples from multiplecancer types [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 1706.

Abstract 1699: Accurate detection of low AF variants relevant to AML by Anchored Multiplex PCR and next generation sequencing

Abstract Introduction: Acute Myeloid Leukemia (AML) is clinically and biologically heterogeneous, requiring the detection of multiple mutations for characterization. For instance, FLT3-ITDs and CEBPA mutations represent important markers in AML, however they are difficult to detect by NGS due to the highly variable nature of ITDs, the high GC content of CEBPA, and the difficulty in mapping repeated sequences to a wild-type reference. Tracking low frequency mutations is also of growing importance. The ability to accurately detect variants at low allele fractions (AFs) using a single test can be used to assess treatment efficacy and potential relapse. Methods: We developed Archer® VariantPlex® myeloid assays based on Anchored Multiplex PCR (AMP™) to detect important mutations in myeloid malignancies. AMP is a target enrichment strategy that uses molecular-barcoded adapters and gene-specific primers for amplification, permitting open-ended capture of DNA fragments from a single end. This approach enables flexible and strand-specific primer design to provide better coverage of ITD-containing and GC-rich regions. We also developed a method to assess SNV sensitivity taking into account both unique coverage depth and noise for single base substitutions. This strategy enables utilization of position-specific detection thresholds and maximizes sensitivity and specificity. We tested this approach using the VariantPlex® Core Myeloid panel, by titrating reference inputs into background normal samples to examine detection of low AF variants. Results: Our assay enables calling of a 30bp FLT3-ITD down to sub-0.05% allele frequencies. Using optimized low AF conditions improves coverage depth, consistency of low AF FLT3-ITD detection, and sensitivity (98.5% of bases are powered to call a true variant at an allele frequency of 3.0% with 1M reads and 200ng of input). We show >1000X unique molecule coverage across the coding region of CEBPA and use this challenging region to visualize the minimum detectable AF (MDAF) at which a variant has a >95% probability of being detected above the noise (95MDAF). Finally, we show consistent single nucleotide variant (SNV), insertion and deletion (indel), and ITD calling at sub-0.5% allele frequencies, and demonstrate the utility of reporting variant-specific MDAFs and normal dataset P-values when analyzing low AF variants. Conclusion: AMP provides NGS-based detection of complex mutation types that are relevant in AML. We demonstrate robust calling of FLT3-ITDs and other variants at low AFs. We also demonstrate full coverage of CEBPA with high depth and low noise such that the 95MDAF predicts confident variant calling at low AFs. This approach is accurate and scalable, enabling simultaneous detection of multiple mutation types across multiple target genes in a single assay. Citation Format: Verity Johnson, Kaitlyn E. Moore, Laura M. Griffin, Aaron Berlin, Abel Licon, Ryan Walters. Accurate detection of low AF variants relevant to AML by Anchored Multiplex PCR and next generation sequencing [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 1699.

Abstract 2474: Automated somatic variant classifier to reduce false positives identified by tumor normal variant callers

Abstract Accurate identification of somatic mutations is key to targeted cancer therapies. However, due to the complexity of next-generation sequencing, there are many variables that give rise to sequencing and alignment artifacts. Most sequencing workflows using tumor normal paired samples rely on the intersection or union of mutations by more than one somatic variant caller such as Mutect2, Strelka, VarScan2, Muse, and SomaticSniper. Yet, they still contain a number of false positive and false negative calls. Therefore, there is a need for a manual review of variants using Integrative Genomics Viewer (IGV) to incorporate information that is difficult to take into account in automated workflows. To alleviate the burden of manual inspection, we present an automated classification method to pre classify detected somatic variants into true or false classes for further manual review. Methods: We created a false positive classifier that utilizes automated IGV visualization of somatic variants. Our somatic workflow involves tumor normal DNA sequencing, BWAMEM alignment, deduplication of reads via Picard, somatic calling using two somatic callers, followed by the union of the detected variants, SnpEff annotation, and extracting nonsilent variants with variant allele fractions greater than 5 percent. We used the data for 1,413 nonsilent variants from a set of 10 tumor normal pairs of various cancer types. The variants were scored by cancer analysts using IGV. Initial false positive rate was 28 percent. We fine tuned Residual Deep Neural Network which utilizes weights from the network with the same architecture trained on ImageNet. Discriminative fine tuning was performed in two stages: first unfreezing the last fully connected layer only, and then unfreezing all the layers and training with differential learning rates. Results and Conclusion: Our method achieves 93 percent overall accuracy. The classifier greatly reduces the false positive rate for the set of variant calls from 28 percent down to 4 percent at the expense of calling 3 percent of true variants false. Previously, deep neural networks have been shown to yield the state of the art performance in multiple domains including image processing, speech, and text processing. In this work, we describe a way to fine tune Residual Neural Network trained on a large image data set to perform false positive variant detection in tumor-normal variant alignment snapshots from IGV. Our general approach that significantly reduces false positive rate of putative variants can be extended to any subset of somatic callers. Citation Format: Alena S. Harley, Corine K. Lau, Eve Shinbrot. Automated somatic variant classifier to reduce false positives identified by tumor normal variant callers [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 2474.

SeqPlus sequencing methodology enables robust whole-genome sequencing, true variant detection, and novel genomic insights from archival esophageal carcinoma FFPE samples.

e13016 Background: Whole-genome sequencing (WGS) of formalin-fixed, paraffin-embedded (FFPE) samples could enable novel insights from archival sample collections, yet robust FFPE WGS is challenged by fragmented DNA, uneven genomic coverage & sequencing artifacts attributed to FFPE fixation. We report our proprietary extraction & library preparation methodology (SeqPlus) with high quality, uniform WGS sequencing performance comparable to that from fresh-frozen samples. Methods: We analyzed 20 paired esophageal carcinoma (EC) samples i.e., primary tumors & matched germline samples to assess SeqPlus performance on 10-15-year-old FFPE tissues, measure variant concordance between WGS and a high-depth sequencing panel (269 genes, 400x coverage) & identify novel genomic features. Results: At a targeted 70x WGS tumor sequencing depth, 93% of the genome was covered by ³ 20 reads, 99% of bases had 10x coverage & average duplicate reads were 31%. We noted similar transition/transversion ratios & mutational spectra as from fresh-frozen EC specimens, suggesting that extraction & library preparation contributes to prior FFPE artifacts. Concordance of tumor-specific SNVs & indels derived from WGS & targeted panel was high at 86%. All 76 targeted panel-detected variants above the WGS limit of detection (mutant allele frequency [MAF] > 10%) were detected by WGS, 2 variants (2 tumors) were detected only by WGS, and 12 variants at MAF ≤ 6% (9 tumors) were only detected by the targeted panel. Tumor WGS yielded SNV, indels & CNV findings beyond variants detected by targeted sequencing. WGS enabled detection of 10.4 putative cancer variants per tumor compared to 12 variants per patient from frozen specimens and a median of 7 (up to 16) cancer-associated variants in genes outside the targeted panel. WGS copy number analysis revealed CCND1, EGFR, TP63, and SOX2amplification, CDKN2A/B deletion and additional unrecognized genomic aberrations. Conclusions: Our study reinforces the utility of high-quality, uniform WGS sequencing of archival FFPE cancer samples with SeqPlus and unlocks the potential for massive-scale retrospective genomic analysis of archived pathology samples with associated clinical & outcomes data.

Overlap graph-based generation of haplotigs for diploids and polyploids.

Haplotype-aware genome assembly plays an important role in genetics, medicine and various other disciplines, yet generation of haplotype-resolved de novo assemblies remains a major challenge. Beyond distinguishing between errors and true sequential variants, one needs to assign the true variants to the different genome copies. Recent work has pointed out that the enormous quantities of traditional NGS read data have been greatly underexploited in terms of haplotig computation so far, which reflects that methodology for reference independent haplotig computation has not yet reached maturity. We present POLYploid genome fitTEr (POLYTE) as a new approach to de novo generation of haplotigs for diploid and polyploid genomes of known ploidy. Our method follows an iterative scheme where in each iteration reads or contigs are joined, based on their interplay in terms of an underlying haplotype-aware overlap graph. Along the iterations, contigs grow while preserving their haplotype identity. Benchmarking experiments on both real and simulated data demonstrate that POLYTE establishes new standards in terms of error-free reconstruction of haplotype-specific sequence. As a consequence, POLYTE outperforms state-of-the-art approaches in various relevant aspects, where advantages become particularly distinct in polyploid settings. POLYTE is freely available as part of the HaploConduct package at https://github.com/HaploConduct/HaploConduct, implemented in Python and C++. Supplementary data are available at Bioinformatics online.