[EMBOSS] [Taverna-users] Bootstrap

Mahmut Uludag uludag at ebi.ac.uk
Mon May 11 15:36:12 UTC 2009


Hi Achille,

> - the fneighbor app fails everytime with errors like this :
>  Error: Distances file found 0 rows, expected 15

This problem looks like have been fixed in CVS version of EMBOSS, as i
get reasonable output when i run CVS version of fneighbor with your
input file. This suggests current patches don't include the fixes I
thought they have. It is my fault that i didn't double check it before
returning you.

> I don't understand the "Standard error stream" explanation! about datafile
> and the "-auto" param.

"-auto" is an emboss parameter which means EMBOSS should try runing an
application without asking further questions but using the
inputs/parameters available on the command line. We will consider
removing it from Soaplab job reports. "Standard error stream" is a term
used by Soaplab meaning "error output of the job", it looks like we can
improve its wording.


> Instead If I use, for exsample, a 48 sequences alignment input I have a
> java heap space error (is it my computer fault?).

This is most probably related to memory used by your Taverna Java
process. I tested your 48 sequence alignment with all 3 workflows
attached, i didn't get any memory errors but i use a quite generous -Xmx
setting in my runme.sh file. You may try increasing the default -Xmx
setting in your taverna runme.sh file. Alternatively, as far as i know
memory used grows each time i run a new workflow in taverna, you may try
closing existing workflow run tabs before running a new workflow. 

> -the timeout limit for the jobs is 60 minutes!? , it is a restriction of
> EBI that is impossible to avoid I suppose!

this was a good enough timeout limit previously, i think we can increase
it, let me discuss this with relevant people, if i execute the same
fproml job outside LSF it takes about 80 minutes which suggests a 2 hour
limit should allow your fproml jobs completed.

> for example a fproml job with 48 seq align is more than 60 minutes long I
> think because after 1 hour it returns a completed job message with empty
> output.

This is a bug in Soaplab (introduced by me), it should report the status
terminated, rather than completed successfully.

Regards,
Mahmut





More information about the EMBOSS mailing list