Closed srynobio closed 9 years ago
Commented by eilbeck on 2008-03-17 20:56 UTC Logged In: YES user_id=742851 Originator: NO
Hi There is already a suggestion in the tracker for a relation 'has_evidence'. Did you see this already? I will add this to SO as soon as we figure out the definition.
I am a bit reluctant to add a very similar relationship but with a qualifier.
How would you feel about merging these requests and working together with the other group on a definition?
--Karen
Commented by jeltje on 2008-03-17 21:01 UTC Logged In: YES user_id=2034688 Originator: YES
Nicole Washington asked me to add this term to distinguish from has_evidence. I feel we need two terms because has_full_evidence means that the gene model is of higher quality than if it just contains has_evidence.
I agree that it's not a very descriptive term, but I can't really think of anything short.
Commented by eilbeck on 2008-03-17 21:29 UTC Logged In: YES user_id=742851 Originator: NO
Hi Jeltje , I'm glad you and Nicole are talking about this. She just mentioned that she is bring others into the discussion too.
I'm trying to work out how to define these terms To start off we I cam e up with these:
has_evidence Feature A has_evidence B if the extent of an instance of B supports the extent of A.
has_full_evidence Feature A has_evidence B if the extent of an instance of B fully supports the extent and internal boundaries of A .
Should these relationships be transitive? I do not think being symmetrical is a good idea and they should not be allowed to be cyclic.
We also need to work out the domain and range. I think for both of these relationships the domain and range should be sequence_feature at the moment. Do you agree?
--Karen
Commented by nlw on 2008-03-18 23:38 UTC Logged In: YES user_id=1526943 Originator: NO
yep, i've been meaning to get back to you about these.
i like your defs. i think it would be good to include an example of each in the comments.
here's what jeltje wants to be able to annotate: "What I want to say is that the transcript is identical to the cluster, so the cluster completely verifies the transcript"
here's an example of gff3 that includes these relationships:
4 . . . . . . . ID=asmbl_110874 4 DMG1 gene 174967 176352 1 - . ID=DMG1-chr4.1.010 4 DMG1 mRNA 174967 176352 1 - . ID=CG31997;Parent=DMG1-chr4.1.010;Parent=asmbl_11087;parental_relationship=has_full_evidence/asmbl_11087 4 DMG1 exon 174967 175215 0 - . Parent=DMG1-chr4.1.010 4 DMG1 exon 175277 175397 0 - . Parent=DMG1-chr4.1.010 4 DMG1 exon 176015 176352 0 - . Parent=DMG1-chr4.1.010 4 DMG1 three_prime_UTR 174967 175028 1 - . Parent=CG31997 4 DMG1 stop_codon 175029 175031 1 - . Parent=CG31997 4 DMG1 CDS 175032 175215 1 - 1 Parent=CG31997 4 DMG1 CDS 175277 175397 1 - 2 Parent=CG31997 4 DMG1 CDS 176015 176153 1 - 0 Parent=CG31997 4 DMG1 start_codon 176151 176153 1 - . Parent=CG31997 4 DMG1 five_prime_UTR 176154 176352 1 - . Parent=CG31997
i think transitive yes, symmetrical probably not.
and this should be a child of has_evidence
Commented by eilbeck on 2008-03-25 22:06 UTC Logged In: YES user_id=742851 Originator: NO
This request was merged with 1911481. A new relationship complete_evidence_for_feature has been created.
Updated by eilbeck on 2008-03-25 22:06 UTC
Reported by jeltje on 2008-03-17 19:48 UTC a new relationship type for "has_full_evidence" so that we can link features to the evidence that fully supports the feature.
for example, if there is 1 (or more) ESTs that have all their splice sites in common with a gene model, then you'd like to say that the mRNA has_full_evidence EST.