Argo File Formats: GFF

GFF (General Feature Format) is a simple tab delimited format for describing genomic features -- simple, except for the fact that there are 3 and a half (and counting) slightly but critically different versions. All consist of 8 required and 1 optional tab delimited fields. They differ solely in how that last optional field is to be interpreted. You'd think that would be a tiny detail, but if argo thinks your file is version 1 and you think it's version 2, chances are you are going to see something weird, useless, and confusing. To avoid such confusion, understand which format you are using and why, and give your files the full format version extensions (example: myfile.gff3 or myfile.gtf2 instead of myfile.gff) so that you remember what you decided and argo doesn't have to guess. Below is a quick list of the different versions with links to more information, an explanation of how field 9 is interpreted, and why you might want to use it.

Version Editable in argo? Grouping support? Attribute support? Field 9 means Pros Cons
GFF1 no yes, one level no Group. All features sharing a value in field 9 will be displayed as subfeatures on a single feature in argo. If all you care about is basic structure of 2 level features, this is the simplest format. There is no way to attach attributes to features.
GFF2 no no yes Attributes (semi-colon delimited). If your features are all single level and all you want to do is be able to attach arbitrary attributes to them, this is the simplest format. There is no way to group features.
GTF2 (stands for Gene Transfer Format) yes yes, two levels (exon to transcript, transcript to gene). Convention for representing start and stop codon. no Grouping by a formatted identifier. Simple way to represent transcripts if all you care about is gene, exon structure, and start/stop codons. You can edit features using the argo UI. There is no way to attach attributes to features.
GFF3 yes yes, theoretically arbitrary, but argo understands up to two levels (exon to transcript, transcript to gene). yes Attributes and grouping using a special attribute. The only format that allows both grouping and attributes. You can do everything you can with GTF2 plus adding attibutes. The spec is a little overambitious. Argo can't display many of the things you can theoretically represent in this format.

Last Updated: Sept 18 2006
Contact: Reinhard Engels