Decoy Proteins Issue

[kusal bellana]

Hi, 

I'm having issues in TPP (Petunia) specifically with decoy proteins appearing in the top results after FDR filtering and being considered correct hits. For example in one of my ipro.prot.xml files I processed through the Petunia pipeline, there are 20 decoy proteins that are at 0.990 probability. However, in the sensitivity/error table for the corresponding file there are no proteins recognized as incorrect at the 0.990 probability. I've included screenshots below for my protein hits after FDR filtering as well as my error/sensitivity table for one of the samples I processed through TPP. Any help with addressing/trouble shooting this issue would be greatly appreciated.

Thanks,

Kusal

[Jimmy Eng]

My guess would be that you did not specify the decoy options when running PeptideProphet. Can you confirm whether you did so or not?  In the "Analyze Peptides" page of the Petunia web interface, there's a checkbox that needs to be checked labeled "Use decoy hits to pin down the negative distribution."  Also enter "DECOY_" in the text box "Known decoy protein names begin with:". 

--
You received this message because you are subscribed to the Google Groups "spctools-discuss" group.
To unsubscribe from this group and stop receiving emails from it, send an email to spctools-discu...@googlegroups.com.
To view this discussion visit https://groups.google.com/d/msgid/spctools-discuss/cfc495ff-76a0-47c1-899e-ed6133f3452cn%40googlegroups.com.

[kusal bellana]

Hi Jimmy,

Here's a screenshot of the analyze peptides page. I include "DECOY_ " in the "known decoy protein names begin with" box each time I run a sample.

I also include the decoy prefix in my params file during the comet search.

 

Best,

Kusal

[Jimmy Eng]

As far as I can tell, you ran the search and tools correctly.  But it looks like there are no IDs in that analysis based on the protein output you included in the original post (irrespective of the protein output list including the decoy entries).  Without seeing your actual data files, I couldn't begin to guess why the decoy entries made it through to the protein output.  Do all of your target-decoy datasets show DECOY_ entries into the protein output list or is it just this one dataset?  From the interact.pep.xml (or whatever your *.pep.xml file is named), can you copy/paste the 5th line of that file that starts with "<peptideprophet_summary" here?  I want to see that the "DECOY=DECOY_" option was actually applied by PeptideProphet.  An example of that line is below:

<peptideprophet_summary version="PeptideProphet  (TPP v7.2.0 Bombogenesis, Build 202411261314-exported (Linux-x86_64))" author="AKeller@ISB" type="unlinked" min_prob="0.05" options=" ACCMASS DECOY=DECOY_ " est_tot_num_correct="27220.9">