Francois JEANMOUGIN wrote:
>> [...Also posted to help at gcg.com...]
> In article <5mfc74$66t at netnews.upenn.edu>,
>ellis at vesicle.dental.upenn.edu (Ellis Golub) writes:
> [...] . . . . .
> > 151 AGNVRSYARMDIGTTHDDYANDVVARAQYYKQHGY............... 185
> > |
> > 1 .................................GASARLLRAAIMGAPGS 17
> > .
> [...]
>> I saw such strange things using GCG9 gap, but had not time
> to work around to see what's happen. I think there is a bug (big one)
> in GCG9 gap. Also, that was with other sequences (but I can't
> remember which ones!!!).
> Could Oxford Molecular (Ooooooops, sorry) GCG look around this
> problem?
The two sequences being aligned here (sw:lyg_ansan and sw:kad3_bovin)
share no significant similarity. Use your most sensitive global
alignment search, or even your most sensitive local alignment search, or
even your most sensitive motif search and I don't think you will find
significant similarity between these two sequences. One of the
sequences encodes a phosphotransferase and the other encodes a
lysozyme. The fact that GAP in GCG version 8 produced a long global
alignment between these two unrelated sequences suggests that the
scoring matrix used by GAP in version 8 to create this alignment may not
have been appropriate. In version 9, GAP produces a minimal carboxy
terminus to amino terminus alignment of these same two sequences. Since
GAP always finds an alignment, no matter how unrelated the sequences
being aligned, this type of minimal alignment between two sequences
indicates to me that they don't really align, at all. So, I think this
strange result is appropriate, in this case at least. If a similar
analysis does not explain other similar results with GAP, BESTFIT, or
PILEUP, please let us know.
--
Irv Edelman, Genetics Computer Group
E-Mail: mailto:edelman at gcg.com Phone: 608 231-5200
WWW: http://www.gcg.com Fax: 608 231-5202
