The decoys are just reversed sequences. I am performing non-specific digestion search, so the chance that spectrum get high-scoring decoy hit is much higher than tryptic peptide search.
I tried the approach I mentioned by doing two searches, one was searching against only foward-sequence, and then other searching against reversed-sequences (Comet search engine). Then in xinteract I specified two input pep.xml files and use -D option to link to the combined FASTA file. But the program stopped and gave error message "NAN probability density detected". Any suggestions? xinteract output: Found 4823 Decoys, and 4820 Non-Decoys Initialising statistical models ... Iterations: ...ERROR: NAN probability density detected. Thanks, Chih-Chiang On Tuesday, April 4, 2017 at 11:20:40 AM UTC-5, David Shteynberg wrote: > > It sounds like a problem with the way your decoys might be getting > generated. How do you generate the decoy? > > On Tue, Apr 4, 2017 at 8:59 AM, Chih-Chiang Tsou <chihchi...@gmail.com > <javascript:>> wrote: > >> Hi David, >> >> Thanks for the prompt response. >> So what I am trying to do is to avoid decoy hit Blocking the possible >> true hit. >> >> The way I perform the analysis is to search data against a FASTA file >> which has reversed sequences appended. Then during PeptideProphet analysis >> I specify the decoy prefix and in the end use the probabilities for >> target/decoy hits to estimate FDR. One potential problem for me is that if >> a spectrum happens to match a decoy as best hit, with a very high score >> let's say XCorr 2.0, and the second hit with Xcorr 1.9, which si actually >> the true hit, would be discarded after PeptideProphet analysis. >> >> If PeptideProphet takes only top hit, could I do two separate searches >> using target-FASTA and Decoy-FASTA and specify two input files in >> PeptideProphet? Any suggestions? >> >> >> Thanks, >> Chih-Chiang >> >> On Tuesday, April 4, 2017 at 9:45:11 AM UTC-5, David Shteynberg wrote: >>> >>> No. It uses the top hit only. >>> >>> On Tue, Apr 4, 2017 at 7:34 AM, Chih-Chiang Tsou <chihchi...@gmail.com> >>> wrote: >>> >>>> Hi >>>> >>>> Is there any option in xinteract to tell PeptideProphet take topN hits >>>> for each spectrum for probability calculation? >>>> >>>> Thanks, >>>> Chih-Chiang >>>> >>>> -- >>>> 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 post to this group, send email to spctools...@googlegroups.com. >>>> Visit this group at https://groups.google.com/group/spctools-discuss. >>>> For more options, visit https://groups.google.com/d/optout. >>>> >>> >>> -- >> 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 <javascript:>. >> To post to this group, send email to spctools...@googlegroups.com >> <javascript:>. >> Visit this group at https://groups.google.com/group/spctools-discuss. >> For more options, visit https://groups.google.com/d/optout. >> > > -- 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-discuss+unsubscr...@googlegroups.com. To post to this group, send email to spctools-discuss@googlegroups.com. Visit this group at https://groups.google.com/group/spctools-discuss. For more options, visit https://groups.google.com/d/optout.