In article <5gpe92$c4c at acebo.sdi.uam.es>,
txomsy at cnb.uam.es (Jose R. Valverde) writes:
> As usual I was too fast in posting. I have continued having problems
> and a more detailed look shew that that was not the only difference. It
> seems that for some reason GCGv9 also computes check numbers differently
> and that's another reason for pileup output not to be recognized properly
> in EGCG8... But some sequences maintain the same checksums (!) and I can't
> see where is the difference. Wait a second! The ones that keep cksum do
> finish in the last column! So that may be it, that last ~ Peter was
> mentioning a while back? Hum...
> OK, I still do not know what makes the new GCG generate those abnormal
checksums and I don't care. I had a look again (despite this terrible cold)
this morning and saw that they really look abnormal.
For one: you can substitute the new checksum by the one in the original
sequence, and then everything works fine again with EGCG8. So the solution
now seems to be editing the failing file to restore the original checsums
from the original sequence files. For now, this seems to affect only those
sequences in files with multiple sequence alignments that happen to end with
a gap character.
But -wiser now- don't take it for granted. I could be coming up with
something new this afternoon. After all, I make a few tests, tell my users
and wait for them to come up with new surprises... Yes I know I may sound
sadistic but I'm not in strong spirits with this cold (maybe I should be *on*
strong spirits?). ;-)
jr
--
Jose R. Valverde
EMBnet/CNB
