Consensus on naming scheme

[GallVp]

Current scheme: AGAT

• Ross's Perl script (gene.t1, gene.t2, etc.)
• mRNA or transcript (GFF3 requires mRNA)
• liftoffID is acceptable (Ross)
• For a gene with multiple different descriptions=differing%20isoform%20descriptions

[GallVp]

Summary of a call with @rosscrowhurst

• NCBI requires that the text in 'product' attribute adheres to a set of rules. These rules keep changing and an automated validation tool is not known.
• JBrowse2 picks the 'description' or 'note' attribute from the gene feature to display as the annotation text. This capability is a high priority for us because many fairGenomes/JBrowse2 users have requested it. Therefore, we should populate the 'description' attribute for the gene features.
• We use eggnogmapper to obtain functional annotations for transcripts. These annotations should be stored both at the transcript and the gene level under the 'description' attribute.
• An experimental feature of pangene is to support multiple isoforms. This might be dropped later. If a gene has multiple isoforms and and they have different functional annotations, this indicates a likely problem with gene prediction. In such a case, the gene level 'description' will be 'differing isoform descriptions'
• To avoid pesky formatting failures, we should use url encoding. Thus, the above description will be stored as: description=differing%20isoform%20descriptions

[GallVp]

Notes:

• @jasonshiller, @rosscrowhurst, @CeciliaDeng Global transcript numbers are confusing. BRAKER uses t1, t2 and that's what we should use. Convention: geneXX.tYY
• @rosscrowhurst A single naming scheme might not work for every case and every user
• Provide transformation tables/web apps for pan-genome

[GallVp]

Completed by https://github.com/PlantandFoodResearch/genepal/releases/tag/0.4.0