• Sign In
  1. Requests

Pecan/walnut

support

View Request

view list print
Pecan/walnut tannous  2017-09-06 08:16
 
Hi,
I have been trying to work with some DIA data using the Walnut/Pecan interface.
However, the output is giving me no peptide matches with 1% FDR. I am testing with DIA data obtained with standard 20mz isolation windows and not DIA from gaz phase fractionnation (GPF) type of acquisition. Does Walnut/Pecan work only with DIA obtained with small isolation windows using GPF or is there something else going on that is causing me to have no peptide matches with 1% FDR?
-Thanks
Abla
 
 
Mike MacCoss responded:  2017-09-06 08:57
Hi Abla,
It will probably require Brian to see your files and your target list of peptides before he can say anything. Another key thing is whether you detect peptides with Skyline using mProphet and a spectrum library?

It is almost certain that narrow isolation windows will give better detection of peptides. Do you have plenty of peptides for the semi-supervised machine learning step to train the SVM on? Going from 10 m/z to 4 m/z isolation windows basically doubles the sensitivity. Maybe start by looking for abundant proteins that should be easy to detect?

The more info you can give the better.

-Mike
 
tannous responded:  2017-09-07 09:36
Hi Mike,
Thank you for your response.

I did not get to the point of importing files into skyline.
I started my workflow in Walnut. I used a protein target list (in a fasta form), of approx 400 of the most abundant proteins and as " background" I used the rat complete fasta. Then I added my mzml file after which walnut processed it and gave me 0 peptides with 1% FDR.
I tried another data set with smaller isolation window of 5mz, it still gave me the same thing.
There is certainly a problem with the training set as it gave me the warning: The mix-max procedure is not well behaved when # targets (388) != # decoys (273). Consider using target-decoy competition (-Y flag).
However, I was only using the default percolator through walnut, I was under the assumption that it would do the search automatically against an appropriate decoy library based on my entries. so I did create a decoy set myself.
I was wondering if you have any recommendations. I could also directly email Brian and send him my data.
Thanks again
Best
Abla.
 
tannous responded:  2017-09-07 09:38
correction*: I did not create the decoy set myself.
 
William responded:  2018-02-28 19:12
How to get Walnut? I searched but cannot find a link for it
 
Mike MacCoss responded:  2018-02-28 23:03
Walnut is a reimplementation of PECAN [https://www.ncbi.nlm.nih.gov/pubmed/28783153]. Walnut was released to attendees of our course for a tutorial. We are just about to submit a manuscript describing a workflow that makes use of Walnut. It will be available soon.

Thanks for your interest,
Mike MacCoss