On Tue, 23 Oct 2001, Cox, Greg wrote: > There's a architectural problem with how locations are represented. The > problem is that strandedness is applied at the feature level, but in some > genbank files applies at the sub-location level. For example, MUSPRCON01 > has the feature: > > gene join(complement(244..1502),L21211.1:1..105, > L21222.1:1..141,L21212.1:1..89,L21213.1:1..101, > L21214.1:1..185,L21215.1:1..225,L21216.1:1..113, > L21217.1:1..116,L21218.1:1..130,L21219.1:1..198, > L21223.1:1..137,L21220.1:1..141,L21221.1:1..664) > /gene="PC4" > > This highlights two things; first, remote feature probably has to be > re-worked. It has no concept of strand, and should. The other is that > there's at least one case where some component parts of a compound location > are on different strands. > > I'm not making any proposals right now, just highlighting the > problem. Let me know if you have brilliant thoughts. > Isn't this sequence representable with ComponentFeatures? At least I use them to effect mappings of this kind in Ragbag. Or have I missed the point completely 9quite possible)?
David Huen _______________________________________________ Biojava-l mailing list - [EMAIL PROTECTED] http://biojava.org/mailman/listinfo/biojava-l
