A standard variation file format for human genome sequences

Abstract

Here we describe the Genome Variation Format (GVF) and the 10Gen dataset. GVF, an extension of Generic Feature Format version 3 (GFF3), is a simple tab-delimited format for DNA variant files, which uses Sequence Ontology to describe genome variation data. The 10Gen dataset, ten human genomes in GVF format, is freely available for community analysis from the Sequence Ontology website and from an Amazon elastic block storage (EBS) snapshot for use in Amazon's EC2 cloud computing environment.

Figure 1

The top-level terms in the Sequence Ontology used in variant annotation. There are 1,792 terms in SO, most of which (1,312) are sequence features. There are 100 terms in the ontology that are kinds of sequence variant, of which the two top level terms are shown, and three sub-types, shown with dashed lines, that demonstrate the detail of these terms. The parts of SO that are used to annotate sequence variation files are sequence alteration to categorize the change (five subtypes shown with dashed lines), sequence feature to annotate the genomic features that the alteration intersects, and sequence variant to annotate the kind of sequence variant with regards to the reference sequence.

Figure 2

An example of a GVF annotation, showing three hypothetical sequence alterations: an SNV, a deletion and a duplication. Lines beginning with '##' specify file-wide pragmas that apply to all or a large portion of the file. Lines are broken over multiple lines and separated by empty lines for presentation in the manuscript, but all data for a given pragma or feature should be contained on one line in a GVF file. A description of the tag-value pairs is given in Table 1.