Dear MS2Rescore developers,
Thank you for your very relevant and helpful tool.
I’ve been integrating MS2Rescore (v3.2.1) into our in-house pipeline and ran into a few issues. I’m sharing them here along with quick fixes. There are likely better solutions, so I’m not submitting code, but I’m happy to help implement improvements if needed.
1. psm_utils dependency
Currently the required version is set to psm_utils>=1.1. For me pip installed v.1.4.1, which had a bug related to generating ScanNr column in .pin file. This issue was fixed in v.1.5, thus I would like to suggest setting dependency as psm_utils>=1.5.
2. Minimal set of required columns in psm file
My choice of input files fell on TSV format, which requires at least two columns peptidoform and spectrum_id. The example input file has many more columns, however it is not clear which of them are necessary. Thus, I had issues due to missing columns precursor_mz and qvalue. The former yield nan values in feature column abs_ms1_error_ppm of .pin file; the later results in 0 identifications in log file (minor, but confuses). Possibly warnings inside code or additional guidelines in documentation can prevent this.
3. Percolator custom arguments
Config file allows to pass custom arguments to Percolator . They are passed to percolator.rescore function, which however later are overwritten. I would suggest to call default arguments and custom arguments dictionary variables differently.
4. Incompatibility of Peptide column in .pin file with Percolator
Percolator expects from Peptide column the formatting Y.n[mod]XXX[mod]XXc[mod].Y, where Y are flanking residues (can be "-" if none) [ref] and terminal modifications have to follow special characters of peptide terminus n/c [ref].
However, MS2Rescore uses psm_utils to write .pin file, which passes peptidoform column as Peptide, adding only flanking dots [ref], no flanking residues. Thus, Percolator fails to read some modified PSMs and this also may be subtle to proper pick-proteins part of algorithm if requested by user (requires knowledge on cleavage) [ref]. Probably it is best to account on this incompatibility in formats inside psm_utils.io.percolator. As a temporary workaround, I modified the .pin file after it was written by psm_utils and before passing it to Percolator (in _write_pin_file). I also restored the original peptidoform in the .pout file after Percolator finished and before it was read by psm_utils (in _update_psm_scores).
5. Percolator fail in subprocess
I am not sure why, but for me Percolator run inside MS2Rescore was failing despite running till the end in command line standalone. Removing capture_output from subprocess command resolved the issue.
If more details are necessary, I'll be happy to provide them.
Dear MS2Rescore developers,
Thank you for your very relevant and helpful tool.
I’ve been integrating MS2Rescore (v3.2.1) into our in-house pipeline and ran into a few issues. I’m sharing them here along with quick fixes. There are likely better solutions, so I’m not submitting code, but I’m happy to help implement improvements if needed.
1. psm_utils dependency
Currently the required version is set to
psm_utils>=1.1. For me pip installed v.1.4.1, which had a bug related to generatingScanNrcolumn in.pinfile. This issue was fixed in v.1.5, thus I would like to suggest setting dependency aspsm_utils>=1.5.2. Minimal set of required columns in psm file
My choice of input files fell on TSV format, which requires at least two columns peptidoform and spectrum_id. The example input file has many more columns, however it is not clear which of them are necessary. Thus, I had issues due to missing columns
precursor_mzandqvalue. The former yield nan values in feature columnabs_ms1_error_ppmof .pin file; the later results in 0 identifications in log file (minor, but confuses). Possibly warnings inside code or additional guidelines in documentation can prevent this.3. Percolator custom arguments
Config file allows to pass custom arguments to Percolator . They are passed to percolator.rescore function, which however later are overwritten. I would suggest to call default arguments and custom arguments dictionary variables differently.
4. Incompatibility of Peptide column in .pin file with Percolator
Percolator expects from
Peptidecolumn the formatting Y.n[mod]XXX[mod]XXc[mod].Y, where Y are flanking residues (can be "-" if none) [ref] and terminal modifications have to follow special characters of peptide terminus n/c [ref].However, MS2Rescore uses psm_utils to write .pin file, which passes
peptidoformcolumn asPeptide, adding only flanking dots [ref], no flanking residues. Thus, Percolator fails to read some modified PSMs and this also may be subtle to proper pick-proteins part of algorithm if requested by user (requires knowledge on cleavage) [ref]. Probably it is best to account on this incompatibility in formats inside psm_utils.io.percolator. As a temporary workaround, I modified the .pin file after it was written by psm_utils and before passing it to Percolator (in _write_pin_file). I also restored the original peptidoform in the .pout file after Percolator finished and before it was read by psm_utils (in _update_psm_scores).5. Percolator fail in subprocess
I am not sure why, but for me Percolator run inside MS2Rescore was failing despite running till the end in command line standalone. Removing
capture_outputfrom subprocess command resolved the issue.If more details are necessary, I'll be happy to provide them.