Gff annotations

[jameshadfield]

Moves feature annotation to a subset of GFF syntax. Closes #187, as future extensions can be added by extending the GFF features we support as needed. This results in changes to the output of augur translate and augur export v2 (see below). The format of mutations is unchanged.

annotations

where	start	end	strand	seqid	type
GFF input	1-based	fully-closed	"+" or "-"
genbank input	1-based	fully-closed	"+" or "-"
augur translate output	1-based	fully-closed	"+" or "-"	included	included
augur export v1 output	0-based	half-open	1 or -1	not included	not included
augur export v2 output	1-based	fully-closed	"+" or "-"	included	included

mutations

nt & aa are both 1-based

[rneher]

with a few minor changes, we can also make augur use annotations like this:

KX369547.1		genome	1	10769			0
KX369547.1		gene	90	456		+	0	gene "CA";
KX369547.1		gene	456	735		+	0	gene "PRO";
KX369547.1		gene	735	960		+	0	gene "MP";
KX369547.1		gene	960	2472		+	0	gene "ENV";
KX369547.1		gene	2472	3528		+	0	gene "NS1";
KX369547.1		gene	3528	4206		+	0	gene "NS2A";
KX369547.1		gene	4206	4596		+	0	gene "NS2B";
KX369547.1		gene	4596	6447		+	0	gene "NS3";
KX369547.1		gene	6447	6828		+	0	gene "NS4A";
KX369547.1		gene	6828	6897		+	0	gene "2K";
KX369547.1		gene	6897	7650		+	0	gene "NS4B";
KX369547.1		gene	7650	10359		+	0	gene "NS5";

This already works for vcf, but it would be quite straight forward to also allow this for fasta alignments. The reference sequence could then be supplied as fasta or simply by name and we don't need to mess around with genbank files anymore. this gff/tsv is much easier to edit.
(it is also straightforward to parse such that we could dump BioGFF and just do it ourselves.)

[emmahodcroft]

I'd def like to keep supporting GenBank though - at least for me that's the fastest way to get annotations for something new. Genbank's GFF export doesn't seem to capture all features. But agree that if you're working on something long-term (or sufficiently in need of editing) this would be an easier format to edit and maintain.

Definitely try out any new GFF parser on the TB GFF & others before dumping BioGFF for them - they can be much less tidy than your example! 😅