Closed ValWood closed 6 years ago
I would be happy for these to be in a separate section, or at the bottom of the list by default, but its trivial and not important...
Do this if quick, otherwise label as future
BTW whan I say "separate section" here, I don't mean a totally separate section.
Tehy are fine in the section they are in, I mean a separate sub-section.
so instead of
protein domains and families blah blah
disorder prediction blah blah
(a slight reordering do a) domains and families are first and b) separate heading for disorder prediction
I'm going to label as "future" as it is very unimportant.
Also MobiDB have gone from InterProt data
MobiDB appears to have replaced Ncoils for low complexity coiled coil/disordered regions.
http://preview.pombase.org/gene/SPAC4H3.11c
old http://www.pombase.org/spombe/result/SPAC4H3.11c#translation
I think these should be listed in their own separate section since they are a bit different from domains and families.
Could can be all proteins which contain "disordered regions".
Hopefully eventually the algorithms will distinguish between the myriad of different types of low-complexity regions in proteins (coiled-coil, loop, intrinsically disordered protein etc). There seems to be a lot of lumping going on....