search

statisticalbiotechnology / triqler

The triqler (TRansparent Identification-Quantification-linked Error Rates)'s source and example code
Apache License 2.0
19 stars 10 forks source link

What other search scores we can use for Triqler? #5

Closed weixiandeng closed 5 years ago

weixiandeng commented 5 years ago

Hi Matt,

For the searchScore column in the input file, except for percolator score, what other search scores we can use here, for example, if we use MaxQuant, can I use the Score column from MSMS.txt file?

Best, Weixian

MatthewThe commented 5 years ago

Hi Weixian,

In theory, you should be able to use any search engine score, just make sure that higher scores mean more confident identifications (or multiply by -1 otherwise). We have, however, not tested this with other scores than percolator's, but I would be very interested in your results.

weixiandeng commented 5 years ago

Hi Mattew,

Thanks for your reply, I'll let you know once I have some results. Since the first glance at your paper, I noticed you did feature extraction at the very begining and mapped to PSM afterwards, so that reminded me of MaxQuant, but it should also work for Skyline-based quantitation pipelines, right? As long as we build the spectrum library with non-filtered PSM, and then extract PSM-intensity pairs which passed mprophet filter. Do you have any test for this kind of pipelines?

MatthewThe commented 5 years ago

I'm not too familiar with the Skyline pipeline so I can't help you too much with that at the moment. However, if you think this would really be a good idea, we can discuss it in some more detail by e-mail.

weixiandeng commented 5 years ago

I failed to find the full decoy search result which means I only found very few decoy from MaxQuant, even searched with decoy concatted database. Do you have any idea if Andomeda gives the intermediate result?

MatthewThe commented 5 years ago

Sorry for the very late reply. As far as I could tell there are no (readable) intermediate files from Andromeda that could provide enough decoy information. Unfortunately, neither the msms.txt nor the allPeptides.txt provided the needed information. I have not checked yet if one could change some search parameters to obtain this, e.g. setting the FDR threshold to 1.0.

MatthewThe commented 5 years ago

As it turns out, there was a file called evidence.txt that seemed to contain sufficient information to run Triqler. I have added a converter for this file to the Triqler package: https://github.com/statisticalbiotechnology/triqler/wiki/Converters

We recommended setting the MaxQuant FDR threshold to 100% for best performance, though it usually also works with the default thresholds of 1% FDR.