Hi David, Many thanks for your explanation - it's now clear to me what's going on.
Cheers, Alastair Am Freitag, 6. März 2020 16:40:25 UTC+1 schrieb David Shteynberg: > > Hello Alastair, > > When you ran PeptideProphet as follows: > > PeptideProphetParser H1_EhN_cm2.pep.xml RT DECOY=XXX_ NONPARAM > > Here is what happened: > > You gave PeptideProphet the names of your decoys but without the flag > DECOYPROBS the decoy hits are all incorrect as far as PeptideProphet is > concerned and they will all get a probability of 0 (unless you enable > DECOYPROBS). > > By default PeptideProphet removes any results with probability less than > 5% unless you set the flag MINPROB=0 (or the flag ZERO). So unless you > apply the flag DECOYPROBS you will not see decoys with non-zero > probabilities. And if you use the flag MINPROB=0 but without DECOYPROBS, > all your decoys will be reported with probability 0, which will not be > helpful if you want to utilize these decoys later. > > If you want to use these decoys later you have to rerun the analysis with > DECOYPROBS, and possibly with MINPROB=0 so you can count the number of > decoy hits at low probabilities to get the decoy rate among random matches, > which you can compare to the decoy rate in you database (they should be > similar.) > > By the way, in my testing I did not find that specifying the same set of > decoys for the iProphet classifier is helpful, when you use iProphet after > PeptideProphet and PeptideProphet already utilized the same decoys. When > the information about the decoys is already used at the PeptideProphet step > it doesn't seem to help the classification at the iProphet step. They > could still be useful as a second measure of FDR, however when the > PeptideProphet DECOYPROBS flag is enabled. > > Hopefully this helps you understand the issue. > > Cheer, > -David > > > > On Fri, Mar 6, 2020 at 5:40 AM 'Alastair Skeffington' via spctools-discuss > <[email protected] <javascript:>> wrote: > >> Hello, >> >> I'm combining some Mascot and comet search results with TPP and have the >> problem that it doesn't seem to be identifying decoys correctly. >> >> I'm running TPP.5.1.0 on ubuntu 16. >> >> The comet and the mascot search are done with decoy search turned off, >> but I added decoys to the database with the prefix "XXX_" >> >> Then I continue to run TPP like this: >> >> InteractParser H1_EhN_cm2.pep.xml H1_ma_EhN.pep.xml H1_co_EhN2.pep.xml >> -D../../Databases/EhuxAllproteins_MCC_decoy.fasta -s -XH1 -Etrypsin >> >> DatabaseParser H1_EhN_cm2.pep.xml >> >> RefreshParser H1_EhN_cm2.pep.xml >> ../../Databases/EhuxAllproteins_MCC_decoy.fasta >> >> PeptideProphetParser H1_EhN_cm2.pep.xml RT DECOY=XXX_ NONPARAM >> >> ProphetModels.pl -i H1_EhN_cm2.pep.xml >> >> tpp_models.pl H1_EhN_cm2.pep.xml >> >> And get the following printed to stdout: >> >> >> ../../../../Databases/EhuxAllproteins_MCC_decoy.fasta,../../Databases/EhuxAllproteins_MCC_decoy.fasta >> - Building Commentz-Walter keyword tree... >> - Searching the tree... >> - Linking duplicate entries... >> - Printing results... >> >> using RT >> Using Decoy Label "XXX_". >> Using non-parametric distributions >> (MASCOT) >> error: -1.0 ion score >> Analyzing H1_EhN_cm2.pep.xml ... >> Parsing search results "/home/ >> mpimp-golm.mpg.de/skeffington/winhome/proteomics2/CAP/CAPdata/mascot/Ehnew/H1 >> >> (MASCOT)"... >> => Found 23168 hits. (0 decoys, 0 excluded) >> => Total so far: 23168 hits. (0 decoys, 0 excluded) >> Parsing search results "/home/ >> mpimp-golm.mpg.de/skeffington/winhome/proteomics2/CAP/CAPdata/H1 >> (Comet)"... >> => Found 21343 hits. (0 decoys, 0 excluded) >> => Total so far: 44511 hits. (0 decoys, 0 excluded) >> File: H1_EhN_cm2.pep.xml >> - in ms run: /home/ >> mpimp-golm.mpg.de/skeffington/winhome/proteomics2/CAP/CAPdata/mascot/Ehnew/H1. >> .. >> - in ms run: /home/ >> mpimp-golm.mpg.de/skeffington/winhome/proteomics2/CAP/CAPdata/H1... >> >> ------------------------------------------------------------------------------- >> TPP DASHBOARD -- started at Fri Mar 6 14:18:51 2020 >> >> ------------------------------------------------------------------------------- >> File H1_EhN_cm2.pep.xml is pepxml >> --> Trying to write file H1_EhN_cm2.pep-MODELS.html >> >> ------------------------------------------------------------------------------- >> Finished at Fri Mar 6 14:18:54 2020 with 0 errors. >> >> ------------------------------------------------------------------------------- >> >> >> So I don't understand why it doesn't find the decoys when they are >> clearly there in the search results, when they are clearly there in the >> .pep.xml file: >> >> <spectrum_query spectrum="H1.27890.27890.4" start_scan="27890" >> end_scan="27890" precursor_neutral_mass="1899.7835" assumed_charge="4" >> index="13563" experiment_label="H1"> >> <search_result> >> <search_hit hit_rank="1" peptide="SWHGEAASKTVDSLPR" peptide_prev_aa="R" >> peptide_next_aa="F" protein="XXX_44165" num_tot_proteins="1" >> num_matched_ions="6" tot_num_ions="60" calc_neutral_pep_mass="1899.7996" >> massdiff="-0.0160" num_tol_term="2" num_missed_cleavages="1" >> is_rejected="0"> >> <modification_info modified_peptide="S[167]WHGEAASKTVDS[167]LPR"> >> <mod_aminoacid_mass position="1" mass="166.998359"/> >> <mod_aminoacid_mass position="13" mass="166.998359"/> >> </modification_info> >> <search_score name="ionscore" value="15.77"/> >> <search_score name="identityscore" value="33.71"/> >> <search_score name="star" value="0"/> >> <search_score name="homologyscore" value="27.00"/> >> <search_score name="expect" value="3.1107"/> >> </search_hit> >> </search_result> >> </spectrum_query> >> >> Any ideas would be much appreciated! >> >> >> -- >> 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 [email protected] <javascript:>. >> To view this discussion on the web visit >> https://groups.google.com/d/msgid/spctools-discuss/90ecb5d5-8e5d-48ea-b19a-ed8fdbc84f59%40googlegroups.com >> >> <https://groups.google.com/d/msgid/spctools-discuss/90ecb5d5-8e5d-48ea-b19a-ed8fdbc84f59%40googlegroups.com?utm_medium=email&utm_source=footer> >> . >> > -- 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 [email protected]. To view this discussion on the web visit https://groups.google.com/d/msgid/spctools-discuss/1af72531-1f74-46d7-bcf8-4bbfd9cc8411%40googlegroups.com.
