AutoPVS1: An automatic classification tool for PVS1 interpretation of null variants

Abstract

Null variants are prevalent within human genome, and their accurate interpretation is critical for clinical management. In 2018, the ClinGen Sequence Variant Interpretation (SVI) Working Group refined the only criterion (PVS1) for pathogenicity in the American College of Medical Genetics and Genomics and the Association for Molecular Pathology (ACMG/AMP) guidelines. The refinement may improve interpretation consistency, but it also brings hurdles to biocurators because of the complicated workflows and multiple bioinformatics sources required. To address these issues, we developed an automatic classification tool called AutoPVS1 to streamline PVS1 interpretation. We assessed the performance of AutoPVS1 using 56 variants manually curated by ClinGen's SVI Working Group and achieved an interpretation concordance of 95% (53/56). A further analysis of 28,586 putative loss-of-function variants by AutoPVS1 demonstrated that at least 27.6% of them do not reach a very strong strength level, with 17.4% based on variant-specific issues and 10.2% on disease mechanism considerations. Moreover, 40.7% (1,918/4,717) of splicing variants were assigned a decreased PVS1 strength level, significantly higher than frameshift and nonsense variants. Our results reinforce the necessity of considering variant-specific issues and disease mechanisms in variant interpretation, and demonstrate that AutoPVS1 is an accurate, reproducible, and reliable tool which facilitates PVS1 interpretation and will thus be of great importance to curators.

Clinical Interpretation of Sequence Variants

“…o Mutalyzer (https://mutalyzer.nl/) (Wildeman, van Ophuizen, den Dunnen, & Taschner, 2008) o VariantValidator (https://variantvalidator.org/) (Freeman, Hart, Gretton, Brookes, & Dalgleish, 2018) o Human Genome Variation Society (HGVS) recommendations for sequence variant nomenclature (https://varnomen.hgvs.org/) (den Dunnen et al, 2016)  Integrated databases: o VarSome (https://varsome.com/) (Kopanos et al, 2019) o VarCards (http://varcards.biols.ac.cn/) (J. Li et al, 2018) o Human Gene Mutation Database (HGMD) (license required for professional version) (http://www.hgmd.org/) (Stenson et al, 2017) o Leiden Open Variation Database (LOVD) (https://www.lovd.nl/) (Fokkema et al, 2011) o ClinGen Evidence Repository (https://erepo.clinicalgenome.org/evrepo/) o Protein Variation Effect Analyzer (PROVEAN) (http://provean.jcvi.org/) (Choi & Chan, 2015) o Mutationtaster (http://mutationtaster.org/) (Schwarz et al, 2014) o PolyPhen-2 (http://genetics.bwh.harvard.edu/pph2/) (Adzhubei et al, 2013) o Human Splicing Finder (http://www.umd.be/HSF3/index.html/) (Desmet et al, 2009) o MaxEntScan (http://hollywood.mit.edu/burgelab/maxent/Xmaxentscan_scoreseq.html) (Yeo & Burge, 2004) o NNSPLICE (http://www.fruitfly.org/seq_tools/splice.html) (Reese et al, 1997) o SpliceAI (https://github.com/Illumina/SpliceAI) (Jaganathan et al, 2019) o subRVIS (http://subrvis.org/) (Gussow et al, 2016) o Pathogenic Variant Enriched Regions (PER) viewer (http://per.broadinstitute.org/) (Pérez-Palma et al, 2019;Perez-Palma et al, 2020) o Constrained Coding Regions (CCRs) Browser (http://www.rebrand.ly/ccrregions/) (Havrilla et al, 2019) o MISCAST (http://miscast.broadinstitute.org/) (Iqbal et al, 2019) o Rare Exome Variant Ensembl Learner (REVEL) (https://sites.google.com/site/revelgenomics/) (Ioannidis et al, 2016) o ClinPred (https://sites.google.com/site/clinpred/) (Alirezaie et al, 2018)  Other useful tools: o alleleFrequencyApp (https://www.cardiodb.org/allelefrequencyapp/) (Whiffin et al, 2017) o Automatic PVS1 interpretation (AutoPVS1) tool (http://autopvs1.genetics.bgi.com/) (Xiang et al, 2019) o MedCalc (https://www.medcalc.org/calc/odds_ratio.php/) o InterVar (http://wintervar.wglab.org/) (Q.…”

“…The SVI suggested using a PVS1 decision tree to determine the LoF effect of a variant in the context of gene structure and molecular etiology, such as alternative splicing or nonsense-mediated mRNA decay (NMD) (Abou Tayoun et al, 2018). An automatic PVS1 interpretation (AutoPVS1) tool (Xiang et al, 2019) has been developed online to facilitate determining the PVS1 strength level for null variants. .…”