{"status":"ok","message-type":"work","message-version":"1.0.0","message":{"indexed":{"date-parts":[[2026,2,6]],"date-time":"2026-02-06T00:28:39Z","timestamp":1770337719215,"version":"3.49.0"},"reference-count":28,"publisher":"Oxford University Press (OUP)","issue":"7","license":[{"start":{"date-parts":[[2019,12,12]],"date-time":"2019-12-12T00:00:00Z","timestamp":1576108800000},"content-version":"vor","delay-in-days":0,"URL":"https:\/\/academic.oup.com\/journals\/pages\/open_access\/funder_policies\/chorus\/standard_publication_model"}],"content-domain":{"domain":[],"crossmark-restriction":false},"short-container-title":[],"published-print":{"date-parts":[[2020,4,1]]},"abstract":"<jats:title>Abstract<\/jats:title>\n               <jats:sec>\n                  <jats:title>Summary<\/jats:title>\n                  <jats:p>We investigate convolutional neural networks (CNNs) for filtering small genomic variants in short-read DNA sequence data. Errors created during sequencing and library preparation make variant calling a difficult task. Encoding the reference genome and aligned reads covering sites of genetic variation as numeric tensors allows us to leverage CNNs for variant filtration. Convolutions over these tensors learn to detect motifs useful for classifying variants. Variant filtering models are trained to classify variants as artifacts or real variation. Visualizing the learned weights of the CNN confirmed it detects familiar DNA motifs known to correlate with real variation, like homopolymers and short tandem repeats (STR). After confirmation of the biological plausibility of the learned features we compared our model to current state-of-the-art filtration methods like Gaussian Mixture Models, Random Forests and CNNs designed for image classification, like DeepVariant. We demonstrate improvements in both sensitivity and precision. The tensor encoding was carefully tailored for processing genomic data, respecting the qualitative differences in structure between DNA and natural images. Ablation tests quantitatively measured the benefits of our tensor encoding strategy. Bayesian hyper-parameter optimization confirmed our notion that architectures designed with DNA data in mind outperform off-the-shelf image classification models. Our cross-generalization analysis identified idiosyncrasies in truth resources pointing to the need for new methods to construct genomic truth data. Our results show that models trained on heterogenous data types and diverse truth resources generalize well to new datasets, negating the need to train separate models for each data type.<\/jats:p>\n               <\/jats:sec>\n               <jats:sec>\n                  <jats:title>Availability and implementation<\/jats:title>\n                  <jats:p>This work is available in the Genome Analysis Toolkit (GATK) with the tool name CNNScoreVariants (https:\/\/github.com\/broadinstitute\/gatk).<\/jats:p>\n               <\/jats:sec>\n               <jats:sec>\n                  <jats:title>Supplementary information<\/jats:title>\n                  <jats:p>Supplementary data are available at Bioinformatics online.<\/jats:p>\n               <\/jats:sec>","DOI":"10.1093\/bioinformatics\/btz901","type":"journal-article","created":{"date-parts":[[2019,12,10]],"date-time":"2019-12-10T20:12:14Z","timestamp":1576008734000},"page":"2060-2067","source":"Crossref","is-referenced-by-count":23,"title":["Lean and deep models for more accurate filtering of SNP and INDEL variant calls"],"prefix":"10.1093","volume":"36","author":[{"given":"Sam","family":"Friedman","sequence":"first","affiliation":[{"name":"Data Sciences Platform, Broad Institute of MIT and Harvard , Cambridge, MA 02142, USA"}]},{"given":"Laura","family":"Gauthier","sequence":"additional","affiliation":[{"name":"Data Sciences Platform, Broad Institute of MIT and Harvard , Cambridge, MA 02142, USA"}]},{"given":"Yossi","family":"Farjoun","sequence":"additional","affiliation":[{"name":"Data Sciences Platform, Broad Institute of MIT and Harvard , Cambridge, MA 02142, USA"}]},{"given":"Eric","family":"Banks","sequence":"additional","affiliation":[{"name":"Data Sciences Platform, Broad Institute of MIT and Harvard , Cambridge, MA 02142, USA"}]}],"member":"286","published-online":{"date-parts":[[2019,12,12]]},"reference":[{"key":"2023062312013674700_btz901-B1","first-page":"402","article-title":"Overfitting in neural nets: backpropagation, conjugate gradient, and early stopping","author":"Caruana","year":"2001","journal-title":"Advances in Neural Information Processing Systems"},{"key":"2023062312013674700_btz901-B2","author":"Chollet","year":"2015"},{"key":"2023062312013674700_btz901-B3","first-page":"22","article-title":"Platinum genomes: a systematic assessment of variant accuracy using a large family pedigree","author":"Eberle","year":"2013","journal-title":"60th Annual Meeting of the American Society of Human Genetics"},{"key":"2023062312013674700_btz901-B4","article-title":"A reference data set of 5.4 million phased human variants validated by genetic inheritance from sequencing a three-generation 17-member pedigree","author":"Eberle","year":"2016","journal-title":"Genome Res"},{"key":"2023062312013674700_btz901-B5","year":"2019"},{"key":"2023062312013674700_btz901-B6","year":"2016"},{"key":"2023062312013674700_btz901-B7","article-title":"Improving neural networks by preventing co-adaptation of feature detectors","author":"Hinton","year":"2012"},{"key":"2023062312013674700_btz901-B8","author":"Ioffe","year":"2015"},{"key":"2023062312013674700_btz901-B9","article-title":"Adam: a method for stochastic optimization","author":"Kingma","year":"2014"},{"key":"2023062312013674700_btz901-B10","doi-asserted-by":"crossref","first-page":"285","DOI":"10.1038\/nature19057","article-title":"Analysis of protein-coding genetic variation in 60, 706 humans","volume":"536","author":"Lek","year":"2016","journal-title":"Nature"},{"key":"2023062312013674700_btz901-B11","article-title":"Aligning sequence reads, clone sequences and assembly contigs with BWA-MEM","author":"Li","year":"2013"},{"key":"2023062312013674700_btz901-B12","author":"Li"},{"key":"2023062312013674700_btz901-B13","first-page":"310458","article-title":"Clairvoyante: a multi-task convolutional deep neural network for variant calling in single molecule sequencing","author":"Luo","year":"2018"},{"key":"2023062312013674700_btz901-B14","doi-asserted-by":"crossref","first-page":"1297","DOI":"10.1101\/gr.107524.110","article-title":"The genome analysis toolkit: a map reduce framework for analyzing next-generation DNA sequencing data","volume":"20","author":"McKenna","year":"2010","journal-title":"Genome Res"},{"key":"2023062312013674700_btz901-B15","first-page":"201178","article-title":"Scaling accurate genetic variant discovery to tens of thousands of samples","author":"Poplin","year":"2017"},{"key":"2023062312013674700_btz901-B16","doi-asserted-by":"crossref","first-page":"983","DOI":"10.1038\/nbt.4235","article-title":"A universal SNP and small-indel variant caller using deep neural networks","volume":"36","author":"Poplin","year":"2018","journal-title":"Nat. Biotechnol"},{"key":"2023062312013674700_btz901-B17","doi-asserted-by":"crossref","first-page":"301","DOI":"10.1109\/JRPROC.1960.287598","article-title":"Perceptron simulation experiments","volume":"48","author":"Rosenblatt","year":"1960","journal-title":"Proc. IRE"},{"key":"2023062312013674700_btz901-B18","first-page":"1","article-title":"Learning representations by back-propagating errors","volume":"5","author":"Rumelhart","year":"1988","journal-title":"Cogn. Model"},{"key":"2023062312013674700_btz901-B19","author":"Sethia","year":"2019"},{"key":"2023062312013674700_btz901-B20","author":"Sethia","year":"2019"},{"key":"2023062312013674700_btz901-B21","first-page":"2951","author":"Snoek","year":"2012"},{"key":"2023062312013674700_btz901-B22","doi-asserted-by":"crossref","first-page":"2997","DOI":"10.1093\/nar\/10.9.2997","article-title":"Use of the perceptron algorithm to distinguish translational initiation sites in E. coli","volume":"10","author":"Stormo","year":"1982","journal-title":"Nucleic Acids Res"},{"key":"2023062312013674700_btz901-B23","author":"Szegedy","year":"2014"},{"key":"2023062312013674700_btz901-B24","first-page":"2818","article-title":"Rethinking the inception architecture for computer vision","author":"Szegedy","year":"2016"},{"key":"2023062312013674700_btz901-B25","author":"Torracinta","year":"2016"},{"key":"2023062312013674700_btz901-B26","doi-asserted-by":"crossref","first-page":"11","DOI":"10.1002\/0471250953.bi1110s43","article-title":"From FASTQ data to high-confidence variant calls: the genome analysis toolkit best practices pipeline","volume":"43","author":"Van der Auwera","year":"2013","journal-title":"Curr. Protoc. Bioinformatics"},{"key":"2023062312013674700_btz901-B27","first-page":"3391","article-title":"Deep sets","author":"Zaheer","year":"2017","journal-title":"Advances in Neural Information Processing Systems"},{"key":"2023062312013674700_btz901-B28","doi-asserted-by":"crossref","first-page":"P31","DOI":"10.1186\/gb-2011-12-s1-p31","article-title":"Genomes in a bottle: creating standard reference materials for genomic variation-why, what and how?","volume":"12","author":"Zook","year":"2011","journal-title":"Genome Biol"}],"container-title":["Bioinformatics"],"original-title":[],"language":"en","link":[{"URL":"http:\/\/academic.oup.com\/bioinformatics\/advance-article-pdf\/doi\/10.1093\/bioinformatics\/btz901\/31617574\/btz901.pdf","content-type":"application\/pdf","content-version":"am","intended-application":"syndication"},{"URL":"https:\/\/academic.oup.com\/bioinformatics\/article-pdf\/36\/7\/2060\/50670091\/bioinformatics_36_7_2060.pdf","content-type":"application\/pdf","content-version":"vor","intended-application":"syndication"},{"URL":"https:\/\/academic.oup.com\/bioinformatics\/article-pdf\/36\/7\/2060\/50670091\/bioinformatics_36_7_2060.pdf","content-type":"unspecified","content-version":"vor","intended-application":"similarity-checking"}],"deposited":{"date-parts":[[2023,6,24]],"date-time":"2023-06-24T21:00:41Z","timestamp":1687640441000},"score":1,"resource":{"primary":{"URL":"https:\/\/academic.oup.com\/bioinformatics\/article\/36\/7\/2060\/5674040"}},"subtitle":[],"editor":[{"given":"Bonnie","family":"Berger","sequence":"additional","affiliation":[]}],"short-title":[],"issued":{"date-parts":[[2019,12,12]]},"references-count":28,"journal-issue":{"issue":"7","published-print":{"date-parts":[[2020,4,1]]}},"URL":"https:\/\/doi.org\/10.1093\/bioinformatics\/btz901","relation":{},"ISSN":["1367-4803","1367-4811"],"issn-type":[{"value":"1367-4803","type":"print"},{"value":"1367-4811","type":"electronic"}],"subject":[],"published-other":{"date-parts":[[2020,4,1]]},"published":{"date-parts":[[2019,12,12]]}}}