annotate genes split across scaffolds about apollo

[selewis]

Absolutely, it was one of the things written about in the proposal. It's
all part of the same problem, folding bits of the genome together when
needed (same technique to be used for intron elision, duplicated regions,
etc.) It all fit under Gregg's general title of genometry. He had it in
IGB, but we never got it into Apollo. I think we have some mockups of the
visualization in the proposal. More next week when I'm back.

-S

On Wed, Oct 22, 2014 at 6:15 PM, Nathan Dunn [email protected]
wrote:

Several of our groups are struggling with annotating genes that are split
across scaffolds. I imagine you guys have discussed the feasibility of this
before… This isn’t really a feature request, but I wanted to put it on your
radar as something that annotators of very fragmented assemblies are
dealing with. We looked a bit into the gff3 and chado support for gene
features split across reference sequences - it appears that gff3 supports
having one feature with a single ID (without having a parent feature
grouping them), but the chado bulk loader doesn’t. There also doesn’t
appear to be a CV term in SO that describes a gene split between sequences
due to assembly fragmentation – which could potentially be used as a parent
feature for these situations. So, I can see where there would be issues
both in figuring out the best way to visualize split genes, as well as how
this should be represented in gff3 or chado.

—
Reply to this email directly or view it on GitHub
#51.

[nathandunn]

The architecture is setup to support this, but almost all of the code I converted from 1.0 assumed that there would always be a single feature location per feature.

[nathandunn]

So . . the issue is that instead of a sequence we are passing in a bookmark and based on that bookmark some location information.

location + bookmark -> reverse-projected location + sequence

So there is a ton of code relating to this in FeatureService . . . mostly it just needs to be generalized for location.

{"track":"{"padding":0, "projection":"None", "referenceTrack":"Official Gene Set v3.2", "sequenceList":[{"name":"Group1.1"},{"name":"GroupUn87"}], "label":"Group1.1::GroupUn87"}:-1..-1","features":[{"location":{"fmin":976735,"fmax":995721,"strand":1},"type":{"cv":{"name":"sequence"},"name":"mRNA"},"name":"GB42183-RA","children":[{"location":{"fmin":995216,"fmax":995721,"strand":1},"type":{"cv":{"name":"sequence"},"name":"exon"}},{"location":{"fmin":976735,"fmax":976888,"strand":1},"type":{"cv":{"name":"sequence"},"name":"exon"}},{"location":{"fmin":992139,"fmax":992559,"strand":1},"type":{"cv":{"name":"sequence"},"name":"exon"}},{"location":{"fmin":992748,"fmax":993041,"strand":1},"type":{"cv":{"name":"sequence"},"name":"exon"}},{"location":{"fmin":993307,"fmax":995721,"strand":1},"type":{"cv":{"name":"sequence"},"name":"exon"}},{"location":{"fmin":976735,"fmax":995216,"strand":1},"type":{"cv":{"name":"sequence"},"name":"CDS"}}]}],"operation":"add_transcript"}

[nathandunn]

Basically a Feature should have multiple feature locations . . and of course those feature locations should be able to go to different reference sequences.

The data structures are setup correctly for this BUT we need to make this something we can support within the code as everywhere the assumption is that there only ever be a single location.

[nathandunn]

In branch: multiscaffold_feature

[nathandunn]

this works except for #1205 , setting the longest ORF. Finishing up there.