cuffdiff problems of Segmentation fault

Question

[xinfengya]

These days I was analyzing my RNASeq data by cufflinks,then I found some problems when I run cuffdiff. The program always died with "Segmental fault" after running some time,my command likes below,

cuffdiff \
  -C pairwise_compare_list \
  -o cuff_diff \
  -p 30 \
  -b genome.fa \
  -u \
  --labels 0.5hDex,0.5hNaCl,2hDex,2hNaCl,0 \
  genome.gtf \
  0.5hDex/abundances.cxb \
  0.5hNaCl/abundances.cxb \
  2hDex/abundances.cxb \
  2hNaCl/abundances.cxb \
  0/abundances.cxb

But cuffdiff ran successfully when I ran with

cuffdiff \
  -C pairwise_compare_list \
  -o cuff_diff \
  -p 30 \
  -b genome.fa \
  -u \
  --labels 0,0.5hDex,0.5hNaCl,2hDex,2hNaCl \
  genome.gtf \
  0/abundances.cxb \
  0.5hDex/abundances.cxb \
  0.5hNaCl/abundances.cxb \
  2hDex/abundances.cxb \
  2hNaCl/abundances.cxb

the only diff is I changed the labels order. I don't known why. Does the sample named by integer can't appear at the end of sample labels?

Even if it ran successfully, when I check results I found that all five comparisons in contrast file pairwise_compare_list were done in genes_exp.diff, but only three comparisons in isoforms_exp.diff. I don't known why, because everything looks well. I have met this problem before, too.

[ivivek_ngs]

Try to allocate the virtual memory and see if it works, might be the bam files are insanely big and the reason it is not being able to work wit the allocated memory. But the label order should be the pointing to the files that are being compared