Sarah, This is Henry Lam -- the original developer of SpectraST. I just replied to David about his problem and I assume you can also see the reply.
But in a nutshell, you need to keep supplying the -M <spectrast.usernmods file> every time you work with the library, including when you try to apply the quality filter, view the library, or search it. SpectraST does not internalize the user-defined modifications (for a good reason, I believe), so it forgets them after your initial command. Henry On Tuesday, December 15, 2015 at 6:05:53 PM UTC+8, Sarah P. wrote: > > Hi David, > > I'm not able to solve your problem, but I can comfort you, you are not the > only one with these kind of problems. Some months ago, I had similar > problems and until today, I could not really solve them. But in the > meantime, I tried a lot of different things. First, I tried to figure out > where exactly my problem is located. And in some points, it seems to be > similar to yours. So, this is what I did: > > I'm also working with a user defined modification. Searches with comet are > successful, just as in your case. In my case, however, the next step was > kind of a quality control as to the respective modification. So I performed > the TPP workflow to get an interact.ipro.pep.xml. Afterwards, I used the > PepXML Viewer to check on the spectral quality of the modified peptides and > to check whether the TPP assigns the same spectrum to the same modified > sequence than other search engines (comparison of original spectra out of > raw files). So far, it works. > > Just like you, I tried to build the spectral library afterwards by > implementing the modification with the user defined modification file you > described. My file spectrast.usermods looks just like yours, only with a > different mass and a different amino acid residue. When I build the > spectral library with the command line of my computer, I also get a > confirmation of a successful building process. But then, I have problems to > work with the library. Normally, I filter the spectral library for > probability and then review the consensus spectra of the modified peptides > in my spectral library. But in this case, when I build a consensus library > with only high probability spectra, there are no modified peptides anymore, > so I cannot perform any searches of data against the library. So I removed > the probability filter and build up a library. But when I try to evaluate > this spectral library to check on the quality of the modified spectra, to > look at the spectra itself or to check whether the modification is set up > correctly, I have some problems: > > · The same spectra I could look at on the level of ipro.pep.xml > (high probability, correct sequence, correct modification) cannot be found > in the splib anymore. > > · In other cases, I find the same spectra now with very low > probabilities. > > · In a third case, the same spectrum is now displayed as a > spectrum containing another phosphorylation or another sequence (comparison > of scan numbers). > > · Most of my reference spectra will be eliminated as soon as I > filter the spectral library for high probability. > > · Just in case I ignore the differences between PepXML Viewer and > Lorikeet Spectrum Viewer and I look at the peptides displayed in the table > of the splib.html, there are some differences between the table and the > spectrum I see when I click at the peptide in the table. For example, in > your case, in the table, I see the sequence DLK[324]DYFTK. When I click on > the spectrum, it will be displayed with the sequence DLK and without the > modification. > > · To be more precise, the corresponding fragment ion series can’t > be matched or assigned as they don’t belong to the sequence displayed > either in the table or next to the spectrum of the Lorikeet Spectrum > Viewer. The precursor mass of the spectrum, however, belongs to the > sequence of the splib.html-table. The fragment masses, interestingly, are > sometimes even higher than the corresponding precursor masses. > > · My next step will be to find out to which sequence the > displayed fragment ion series could belong, but even if I can figure this > out, I don’t know how this could help. > > · Additionally, when I compare just the raw spectrum (the > spectrum with the same scan number than the spectrum the library took for > the display. I extracted the necessary information directly out of the xml > files.) with the results of other programs, the other softwares will find > another sequence belonging to this raw spectrum. I did not work with a > consensus library when comparing different softwares of course. > > · To complete the view, I just searched for any modified peptides > of the interact.ipro.pep.xml in the spectra of the splib.html, but I found > none. I clicked at every single Lib-ID that corresponds to a modified > peptide and the problem is always the same. > > I repeated these tests with different data sets, different versions of the > TPP and of spectrast, different formats of spectrast.usermods, different > locations of the TPP (server based and local) and also different user > defined modifications. The problem is always the same. > > > > So, I’m sorry that I cannot really help you, but maybe I can guess that > there is no problem with your data set or the way you perform your search > against the library, but only with the user defined modification in the > spectral library. > > > > Is there anyone else out there having similar problems or maybe already a > solution for it? > > > > Thanks in advance! > ... -- 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 post to this group, send email to [email protected]. Visit this group at https://groups.google.com/group/spctools-discuss. For more options, visit https://groups.google.com/d/optout.
