dbNSFP v4: a comprehensive database of transcript-specific functional predictions and annotations for human nonsynonymous and splice-site SNVs.
Liu X, Li C, Mou C, Dong Y, Tu Y. · 2020
Read dbNSFP v4: a comprehensive database of transcript-specific functional predictions and annotations for human nonsynonymous and splice-site SNVs. by Liu X, Li C, Mou C, Dong Y, Tu Y. (2020) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/europmc-pmc7709417.txt.
Introduction
Liu and co-authors update a database that judges human gene variants. The database, dbNSFP version 4, covers single-letter variants that change a protein or a splice site. Such data pour in from exome sequencing in disease studies. Picking the harmful few from the harmless mass needs strong filters.
The database holds predictions for 84,013,093 possible variants in the human genome. It compiles 36 harm scores, including 12 that are specific to each transcript. Gene-level and variant-level notes sit alongside the scores. The resource is free online, with a download and a web service. The authors say it speeds the hunt for disease-causing variants.
Catalog record
| CREATOR | Liu X, Li C, Mou C, Dong Y, Tu Y. |
|---|---|
| YEAR | 2020 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 27 KB |
| LICENSE | CC BY 4.0 |
| RIGHTS | CC BY |
| SOURCE | Europe PMC |
Browse the full library · About the Tower of Babel
Full text
The opening of the work, from the library's hosted text:
dbNSFP v4: a comprehensive database of transcript-specific functional predictions and annotations for human nonsynonymous and splice-site SNVs By Xiaoming Liu, Chang Li, Chengcheng Mou, Yibo Dong, Yicheng Tu Published in: Genome Medicine DOI: 10.1186/s13073-020-00803-9 ABSTRACT Whole exome sequencing has been increasingly used in human disease studies. Prioritization based on appropriate functional annotations has been used as an indispensable step to select candidate variants. Here we present the latest updates to dbNSFP (version 4.1), a database designed to facilitate this step by providing deleteriousness prediction and functional annotation for all potential nonsynonymous and splice-site SNVs (a total of 84,013,093) in the human genome. The current version compiled 36 deleteriousness prediction scores, including 12 transcript-specific scores, and other variant and gene-level functional annotations. The database is available at http://database.liulab.science/dbNSFP with a downloadable version and a web-service. Supplementary information The online version contains supplementary material available at 10.1186/s13073-020-00803-9. BACKGROUND Whole-exome sequencing (WES) and whole-genome sequencing (WGS) have been increasingly used in human disease studies in the research and clinical setting [1–3]. As a result, we witness a tsunami of DNA sequence data from both healthy individuals and those with Mendelian or complex diseases. Identifying variants that cause diseases or are associated with disease risks from a large number of DNA variants identified in sequencing requires an excessive amount of time and effort. To accomplish this daunting task, investigators have relied on functional annotations to filter or prioritize variants based on our current knowledge or prediction models. In more detail, functional annotations can be separated into general annotation and functional prediction: the former provides qualitative or descriptive annotation of a variant indirectly related to its potential function, such as whether the variant is a nonsynonymous SNV; the latter typically provides direct quantitative or yes-or-no deleteriousness prediction of the variant based on a statistical model. Fast and comprehensive functional annotation tools will become even more critical in the near future because of three intertwined ongoing trends: the decreasing cost of DNA sequencing, the development and practice of precision medicine [4], and the adaptation of WES and WGS in small labs [5]. There have been several annotation tools available for large-scale DNA sequence data, such as UCSC Genome Browser’s Variant Annotation Integrator [6], Ensembl’s Variant Effect Predictor (VEP) [7], ANNOVAR [8], and SnpEff [9]. Most of these focused on general annotations based on given gene models. Although gene-model based annotations are handy, there are other important functional annotation resources used by medical geneticists and genetic epidemiologists, including functional prediction of variants, conservation information, predicted promoters, enhancers, and epigenomic markers, among others. Another challenge faced by the investigators is that different gene-model-based annotation tools all have their advantages and disadvantages, and the results sometimes do not agree with each other [10]. Therefore, it has been suggested to obtain annotation from tools across multiple databases for a complete interpretation of the variants. Previously, we developed dbNSFP version 1 [11], 2 [12], and 3 [13] to provide a “one-stop-shop” for functional annotations for non-synonymous SNVs (nsSNVs) and splice site SNVs (ssSNVs), top candidate variant types for Mendelian diseases. It collected all possible nsSNVs and ssSNVs based on human reference sequences and multiple deleteriousness predictions and annotations for each SNV. Here we report the major updates of dbNSFP since version 3.0 to the current version 4.1. The core SNVs have been rebuilt based on human reference sequence version hg38 and GENCODE version 29 [14]. Compared to version 3.0 [13], dbNSFP v4.1 added 18 deleteriousness prediction scores (BayesDel_addAF and BayesDel_noAF [15], CADD_hg19 [16], ClinPred [17], DEOGEN2 [18], Eigen and Eigen PC [19], FATHMM-XF [20], GenoCanyon [21], LINSIGHT [22], LIST-S2 [23], M-CAP [24], MPC [25], MutPred [26], MVP [27], PrimateAI [28], REVEL [29], SIFT4G [30]), one score for loss of function prediction (ALoFT [31]), and three conservation scores (phyloP17way_primate [32], phastCons17way_primate [33], bStatistic [34]), making the total number of prediction scores to 46 (Additional file 1: Table S1). Many other functional annotation resources have been added or updated. In addition to the previously supported query of two attached databases, dbscSNV [35] and SPIDEX [36], for predicting splice interrupting SNVs, the companion query program for the downloadable version added support for querying SpliceAI, a third-party database for predicting splice site gain and loss [37], and dbMTS, a comprehensive database for microRNA target site SNVs and their functional predictions [38]. More importantly, much effort has been made to increase further the usability of the functional annotations, including (1) making functional predictions transcript-specific whenever possible, (2) providing transcript annotations to help to choose appropriate transcript from multiple isoforms for each gene, (3) providing HGVS (Human Genome Variation Society) c. and p. presentations of the SNVs to facilitate the query of candidate mutations reported in medical genetics literatures, and (4) providing graphic user interface for querying downloadable version as well as web-service for researchers with minimum bioinformatics training. CONSTRUCTION AND CONTENT We rebuilt the list of all potential nonsynonymous and splice-site SNVs based on the GENCODE gene model version 29 (Ensembl version 94) with human reference sequence GRCh38. Only transcripts with complete protein-coding annotations were included. A total of 81,782,923 nsSNVs and 2,230,170 ssSNVs were collected in the database (Additional file 2: Table S2). The corresponding chromosomal positions of the SNVs based on human reference sequences hg19 and hg18 were obtained via the liftover tool [39] (Additional file 2: Table S2). Accurate protein ID mapping between GENCODE/Ensembl and Uniprot [40] was obtained via a comprehensive protein sequence matching between all the proteins in GENCODE/Ensembl and those of the Uniprot database. To facilitate the choice of the appropriate transcript(s) for each gene, we collected transcript quality measures including APPRIS [41], transcript support level (TSL), GENCODE Basic, and Ensembl canonical transcripts were obtained from the Ensembl Biomart [42] and Variant Effect Predictor (VEP). HGVS c. and p. presentations by ANNOVAR, snpEff, and VEP for each nsSNV and ssSNV were obtained via the WGSA (WGS Annotator) pipeline [43]. As a core content of dbNSFP, 36 deleteriousness prediction scores, nine conservation scores, and one loss of function score for each nsSNV or ssSNV were compiled (see Additional file 1: Table S1 for a summary). Among them, 13 scores are transcript-specific (ALoFT, DEOGEN2, FATHMM [44], LIST-S2, MPC, MutationAssessor [45], MVP, Polyphen2 HDIV and Polyphen2 HVAR [46], PROVEAN [47], SIFT [48], SIFT4G, VEST4 [49]). The full list of annotation resources and the description of all columns in dbNSFP can be found at http://database.liulab.science/dbNSFP. UTILITY AND DISCUSSION Query utility dbNSFP v4.1 can be accessed as either a downloadable and standalone version, or as a web-service at http://database.liulab.science/dbNSFP. The standalone version is suitable for a large-scale query, such as quickly identifying nsSNVs and ssSNVs from exome sequencing studies. As no internet connection is required, maximum speed and security can be achieved. The query can be conducted via the companion Java program, which supports both the command-line and graphic user interface (GUI). The query term can be either a genomic position (chromosome, position), an SNV (chromosome, position, reference allele, alternative allele), an amino acid (AA) change (chromosome, position, reference allele, alternative allele, reference AA, alternative AA), a dbSNP ID (rs number), an HGVS c. or p. presentation of a mutation, or a gene name or ID. The companion Java program also supports searching attached databases