#178 Seg fault in p-value branch

[William S Noble]

I am getting a seg fault from tide-search using p-values. All of the relevant files are at

http://noble.gs.washington.edu/~wnoble/proj/crux-projects/2014multiple-testing/results/bill/2014-06-17version2/

In particular, the file runall.lysine.washington.edu.out shows the stdout from the runall script, including the problematic command line and resulting seg fault:

/net/noble/vol1/home/noble/proj/crux/branches/crux-pvalue/src/c/crux tide-search --precursor-window 50 --parameter-file crux.prm --output-dir plasmo.tide-search ../../../data/spectra/v07551_UofF_malaria_TMT_13.ms2 plasmo
INFO: Beginning tide-search.
INFO: Writing results to output directory 'plasmo.tide-search'.
INFO: CPU: lysine.gs.washington.edu
INFO: Wed Jul 9 09:56:05 PDT 2014
INFO: Running tide-search...
INFO: Reading index plasmo
INFO: Reading spectra file ../../../data/spectra/v07551_UofF_malaria_TMT_13.ms2
INFO: Converting ../../../data/spectra/v07551_UofF_malaria_TMT_13.ms2 to spectrumrecords format
INFO: Sorting spectra
INFO: Running search
./runall: line 217: 28217 Segmentation fault (core dumped) $CRUX tide-search --precursor-window 50 --parameter-file $parameters --output-dir $outDir $MS2 $db

[Attila Kertesz-Farkas]

I am not 100% sure, but I think the p-value branch is not accepting bin-width and offset parameters different from default param settings. In your parameter file I saw:
mz-bin-width=0.03
mz-bin-offset=0.5

a.

[Jeff Howbert]

After looking at the parameter file I had the same suspicions as Attila. He is correct. The current crux-pvalue branch does not support variable mz-bin-width.

Does trunk now include support for variable bin width? I could try merging trunk into the crux-pvalue branch again. However, without some significant re-engineering of the dynamic programming, mz-bin-width = 0.03 will take an intolerably long time to run.

[William S Noble]

The smaller bin-width is not critical for my experiments. I will just
re-do this with the default bin width. You should definitely ensure,
though, that the program fails gracefully rather than seg faulting when a
non-standard bin width is requested.

Bill

On Wed, Jul 9, 2014 at 12:47 PM, Jeff Howbert howbert@users.sf.net wrote:

After looking at the parameter file I had the same suspicions as Attila.
He is correct. The current crux-pvalue branch does not support variable
mz-bin-width.

Does trunk now include support for variable bin width? I could try merging
trunk into the crux-pvalue branch again. However, without some significant
re-engineering of the dynamic programming, mz-bin-width = 0.03 will take an
intolerably long time to run.

---

• [issues:#178] http://sourceforge.net/p/cruxtoolkit/issues/178/ Seg
  fault in p-value branch*

Status: open
Milestone: p-values
Labels: high priority
Created: Wed Jul 09, 2014 05:24 PM UTC by William S Noble
Last Updated: Wed Jul 09, 2014 07:33 PM UTC
Owner: Jeff Howbert

I am getting a seg fault from tide-search using p-values. All of the
relevant files are at

http://noble.gs.washington.edu/~wnoble/proj/crux-projects/2014multiple-testing/results/bill/2014-06-17version2/

In particular, the file runall.lysine.washington.edu.out shows the stdout
from the runall script, including the problematic command line and
resulting seg fault:

/net/noble/vol1/home/noble/proj/crux/branches/crux-pvalue/src/c/crux
tide-search --precursor-window 50 --parameter-file crux.prm --output-dir
plasmo.tide-search ../../../data/spectra/v07551_UofF_malaria_TMT_13.ms2
plasmo
INFO: Beginning tide-search.
INFO: Writing results to output directory 'plasmo.tide-search'.
INFO: CPU: lysine.gs.washington.edu
INFO: Wed Jul 9 09:56:05 PDT 2014
INFO: Running tide-search...
INFO: Reading index plasmo
INFO: Reading spectra file
../../../data/spectra/v07551_UofF_malaria_TMT_13.ms2
INFO: Converting ../../../data/spectra/v07551_UofF_malaria_TMT_13.ms2 to
spectrumrecords format
INFO: Sorting spectra
INFO: Running search
./runall: line 217: 28217 Segmentation fault (core dumped) $CRUX
tide-search --precursor-window 50 --parameter-file $parameters --output-dir
$outDir $MS2 $db

---

Sent from sourceforge.net because you indicated interest in
https://sourceforge.net/p/cruxtoolkit/issues/178/

To unsubscribe from further messages, please visit
https://sourceforge.net/auth/subscriptions/

[Attila Kertesz-Farkas]

Yes, crux-tide-trunk supports now variable bin-width and offset.

[William S Noble]

Jeff reports that he has fixed this by preventing people from using small bin widths in conjunction with the p-value option.