Protein/peptide extractions from multiple runs

support
Protein/peptide extractions from multiple runs zainab noor  2018-11-28
 

Dear Skyline Team,

This may be a very basic question but what should I do to see peptides which may be present in any two replicates if not all after I apply the mProphet and q-value cutoff. For example, a peptide which is present in 2 out of three replicates only with 99% FDR. Currently, I am seeing the peptides which are found in all the replicates with 99% FDR after I integrate the peaks using mProphet model.

Thanks for all the help.

Best,
Zainab

 
 
Brendan MacLean responded:  2018-11-28

Hi Zainab,
Not sure I follow what you mean. Where exactly are you seeing the peptides? How are you applying the q-value cut-off? I assume you mean 1% FDR?

--Brendan

 
zainab noor responded:  2018-11-28
Yeah sorry I meant 99% FDR and I’m talking about Skyline.
When I extract the chromatogram from DIA runs into the Skyline, use decoys to creat a mProphet model, use 0.01 q value and integrate the peaks, there can be some peptides which may be found in any 2 runs and not in all runs/replicates. Does Skyline pick only those peptides which pass the qvalue threshold in all the replicates or does it also pick the ones which pass the qvalue threshold in any of the replicates?

I hope I explained it better.

Thanks.

Best,
Zainab
 
Brendan MacLean responded:  2018-11-28
Can you post a screenshot just so I am sure? Are you referring to the "Only integrate significant q values" option in the "Reintegrate" form? You are setting the "Q value cutoff" to 0.01? (or 1% FDR)

That is a peak-level (or query level) cutoff, with the impact that if the peak-level (i.e. peptidoform, chromatogram peak group) q value is greater than 0.01, then the peak is not integrated and all peak statistics are given #N/A values.

There is no setting I know of in Skyline that would allow you to disallow integration based on the run-count where detections occurred, but there may be in the next version of Skyline. We are working on adding both peptide-level and protein-level q values which can be used to exclude results where a peptide or protein is not detected below a certain q value cut-off (or corresponding FDR) even though some of its peaks may still have a low enough peak-level q value (which makes those peak detections more likely false-positives). See the following paper for a discussion of these concepts:

https://www.ncbi.nlm.nih.gov/pubmed/28825704

Once we have these added q value types, we intend to also add a new "Detection" tab to the Refine > Advanced form, which will also contain filters for peak-level detection by number of runs, as you seem to be suggesting.

Not there yet, though. If I have misunderstood what you are doing or asking for, please reply with new details.

--Brendan
 
zainab noor responded:  2018-11-29
Hi Brendan,

Thanks for the reply. Got your point. Yes, I was referring to "Only integrate significant q values" option in the "Reintegrate" form.

I'll have a look at this paper.

Best,
Zainab