smith-chem-wisc / MetaMorpheus

Proteomics search software with integrated calibration, PTM discovery, bottom-up, top-down and LFQ capabilities
MIT License
91 stars 46 forks source link

Issue with Peptide Assignment and Protein Grouping in Software v.1.0.2 #2410

Open AmirWIS opened 2 months ago

AmirWIS commented 2 months ago

I am encountering an issue with software v.1.0.2. It reports that 25 identified peptides were assigned to a protein, 4 of which are unique. I expected to see which peptides are shared and which are unique, but I can't find that information. Additionally, although some peptides are shared, only one (main?) protein is listed in the protein group and also in the "AllPeptides" and "AllQuantifiedPeptides" outputs. When manually reviewing the identified peptides against the Human database, I found that some peptides are shared with other proteins, which are not reported in either the Proteins or Peptides lists. Please assist. Thanks, Amir

trishorts commented 2 months ago

The unique and shared peptides are in the AllProteinGroups.tsv file image

trishorts commented 2 months ago

We use a standard protein parsimony approach where we report the fewest proteins that explain all of the peptides. In this case, since there is a unique peptide, that goes on the protein list by itself. This protein then swallows up all shared peptides unless there is some other protein which also has a unique peptide. I believe your output is correct, but I could look at it if you like. Happy to discuss if you'd like to have a zoom.

AmirWIS commented 2 months ago

@trishorts Hi, you can download the raw files using this link: https://owncloud.incpm.weizmann.ac.il/index.php/s/w6DXPJ4MSE78q4n The protein of interest is P07910. Although the report indicates both shared and unique peptides, it doesn’t specify the other protein(s) in the group, nor does it show them in the peptide results. Thanks, Amir

trishorts commented 2 months ago

i am searching these files now

trishorts commented 2 months ago

That protein has 7 unique peptides. Even though there are 23 shared peptides (with other proteins), no other proteins are listed because none are required to explain all the peptides. If you "turn of protein parsimony/inference" in the search, then I think you will see all the possible proteins for each peptide. But there will not be a protein group output. I'll try tomorrow. I have to go home now today.

trishorts commented 2 months ago

these are the peptide identifications that I get when I turn of protein parsimony. some of the peptides have multiple accessions. perhaps this is what you seek. noProteinInference.txt