This post is part 3 of a series on file formats, written for the 2017 UK-KBRIN Essentials of Next Generation Sequencing Workshop at the University of Kentucky. The conference website is hosted here.

General transfer format (GTF), also known as General Feature Format (GFF) 2.0, is the format for transcripts in exercise 4, RNAseq. For more details, please see the ensembl guide to GFF.

GTF/GFF2.0 is fairly straightforward. Understand that is can describe a variety of sequence features, not just transcripts. All components of a gene's structure, such as introns, exons, and protein CDS, can be described with GTF using the correct feature type (column 3).

Format

Each feature takes up one line, with 9 columns per line (plus optional track definition lines).

Columns

1. seqname Chromosome or scaffold name.
2. source Database, project, or program name.
3. feature Feature type (eg Gene)
4. start Start position
5. end End position
6. score
7. strand + (forward) or - (reverse)
8. frame 0, 1, 2
9. attribute Tag "value" ; tag "value"

1 transcribed_unprocessed_pseudogene  gene        11869 14409 . + . gene_id "ENSG00000223972"; gene_name "DDX11L1"; gene_source "havana"; gene_biotype "transcribed_unprocessed_pseudogene"; 

In the above sample GTF file from ensembl, for example, the name is (chromosome) 1, the source is transcribed_unprocessed_pseudogene, the feature type is gene, the start-end is 11869-14409, there is no applicable score (.), it's on the forward (+) strand, there is no frame data, and there are four attributes. The first attribute, for example, is gene_name with the value DDX11L1.

Track lines

The optional track lines start with the word 'track', followed by space-seperated key=value pairs.