Hi, and thank you for FDRBench. I have a question about using protein-level shuffled entrapment with a non-specific immunopeptidomics search.
Setup
- FDRBench 1.2.0,
-level protein -fold 10 -fix_nc c -check, default enzyme (Trypsin, no P rule)
- Search: MSFragger with non-specific digestion (
num_enzyme_termini = 0), 7-25 aa, HLA class I data
- I checked every accepted entrapment-target peptide for an exact match to the real target sequences in the same database
What I see
At 1% FDR, 99 of the 157 unique accepted entrapment-target peptides (63%) also occur exactly in a real target protein. 120 of the 157 span a junction between two shuffled tryptic blocks, and none come from a shuffle that left the sequence unchanged.
That makes sense for a tryptic search, where each shuffled block is what gets matched. A non-specific search can match any 7-25-residue window, including windows that cross block boundaries, so those real sequences can come back through the entrapment proteins.
Questions
- For non-specific searches, is
-enzyme 0 the intended setting for protein-level entrapment, or would you recommend a different approach?
- Would a window-level check (every peptide in the search's length range against the target set) be in scope for FDRBench, or is that better done downstream?
Happy to share the scripts I used for the check.
Hi, and thank you for FDRBench. I have a question about using protein-level shuffled entrapment with a non-specific immunopeptidomics search.
Setup
-level protein -fold 10 -fix_nc c -check, default enzyme (Trypsin, no P rule)num_enzyme_termini = 0), 7-25 aa, HLA class I dataWhat I see
At 1% FDR, 99 of the 157 unique accepted entrapment-target peptides (63%) also occur exactly in a real target protein. 120 of the 157 span a junction between two shuffled tryptic blocks, and none come from a shuffle that left the sequence unchanged.
That makes sense for a tryptic search, where each shuffled block is what gets matched. A non-specific search can match any 7-25-residue window, including windows that cross block boundaries, so those real sequences can come back through the entrapment proteins.
Questions
-enzyme 0the intended setting for protein-level entrapment, or would you recommend a different approach?Happy to share the scripts I used for the check.