Comet Fragment Ion indexing crashes

[wille...@gmail.com]

Hey,

I am trying to implement the Comet fragment ion indexing to speed up our immunopeptidomics searches. I currently get this issue:

Comet version "2025.01 rev. 1 (4181df6)"

 Creating plain peptide/protein index file for fragment ion indexing:
 - parse peptides from database ... 100%
 - remove duplicate peptides
 - write peptides/proteins to file
 - mods: M, n, C, E, Q
 - initializing combinations (peptide length 20, max mods 1, total combinations 20)
 - 125996860 modifiable peptides; 85906 unique modifiable sequences
   - get modification combinations ... 0 sec
 - done. ../../../Proteomics/CMB/CMB-1580/UP000005640_9606_07082024.fasta.idx (147101155 peptides)

 - read precursors ... 0 sec
Segmentation fault (core dumped)

So it is an unspecific search where I set length 7 to 20, one variable modification, and I already tried setting fragindex_max_fragmentmass = 1500. I am running this on a 500 Gb RAM server and I do not see this spiking, just crashing immediately on the read precursors step. Also there is sufficient space for idx creation.

Thanks,

Patrick

[Jimmy Eng]

Not strictly necessary as I can make up some search settings but can you send me (via email or attached here) your comet.params?  That'll remove a variable as I attempt to debug this.

[Jimmy Eng]

I will need your comet.params.  I've run a couple of tests here (no-enzyme, 7 to 20 length, the five variable mods) and amazingly the searches completed.  Your search is dying after "read precursors" so there's some issue associated with the precursor parsing (although not likely as that step seems to have completed) or it's associated with the following step which is reading in the .idx file.  Anyways, the next step is for me to try running the fragment ion indexing through using your exact params settings to see if I can replicate the segfault you're getting. 

[wille...@gmail.com]

Hey Jimmy,

Thanks for looking into this. I generated a novel parameter file but indeed did some adaptations, so perhaps that brings the error.
It's attached here in this post.

Thanks,

Patrick

[Jimmy Eng]

Patrick,

I'm not able to reproduce the issue you're observing using your search parameters.  My test searches run through fine; run output attached.  I guess the next steps are:

• Regenerate your .idx file and try again.  If you've already done this step and it still invokes the same error, ignore.
• Make your human fasta , corresponding .idx file, and spectrum query file available for me to download.
  

Jimmy

 Comet version "2025.01 rev. 1 (4181df6)"

 Creating plain peptide/protein index file for fragment ion indexing:
 - parse peptides from database ... 100%
 - remove duplicate peptides
 - write peptides/proteins to file
 - mods: M, n, C, E, Q
 - initializing combinations (peptide length 20, max mods 1, total combinations 20)

 - 143019611 modifiable peptides; 90249 unique modifiable sequences

   - get modification combinations ... 0 sec

 - done. HUMAN.fasta.20250131.idx (166956179 peptides)

 Comet version "2025.01 rev. 1 (4181df6)"

 - read precursors ... 9 sec
 - read .idx ... 1 min 47 sec
 - generate fragment index
   - count fragment index vector sizes ... 12 sec
   - reserve memory ... 29 sec
   - populating index ... 9 min 45 sec
   - sorting fragment mass bins by peptide mass ... 1 min 0 sec
   - total # of entries in the fragment index: 18953881593 (max 8320)
 - searching "20231228_Hela_QE_01" ... 5 sec
 - searching "20231228_Hela_QE_02" ... 7 sec
 - searching "20240226_Lu_100ng_Hela_OTMS2_02" ... 52 sec
 - done.

[wille...@gmail.com]

Hey Jimmy,

I managed to fix the error, it seemed to some sort of permission error and if I placed the FASTA locally and its index generation there it went just fine as your output.
Sorry for the waste of time,

Patrick