Aengus Stewart <aengus.stewart at icrf.icnet.uk> writes:
>> I am getting a core dump - floating point exception when runnning 9.0
> FASTA against the Eukaryotic Promoter Database (EPD) from Phillip
> Bucher.
>> I am building the GCG formatted files from the file that includes
> sequence data from -499 to +100 around the promoter.
>> I have tested the formatted files with stringsearch, fetch and
> wordsearch as well. All of these work.
>> I went back and dug out 8.1 and tried the old FASTA. This also works.
>> Anybody else come across this problem? the only thing I can think of
> that could be a problem is the high number of Ns in the sequences.
> However as 8.1 handled it why doesnt 9.0?
>
The reason that gcg's fasta is dumping core is an inability to fit the
similarity scores to estimate the statistics. It has little to do
with N's.
The solution should be to turn off the estimation procedure (or use my
FASTA, which is probably more robust to these problems).
Look for a -z=-1, -z=0, or some command line parameter to deactivate
the statistical estimates.
Bill Pearson
