Hi David, Thanks for your reply, that's very helpful.
I've now tried including an equal number of decoys (reversed) in my database, processing the same data through peptide prophet, iprophet and protein prophet. My number of identified proteins has gone down, but the PSMs are much more believable (judging from the spectra). As an aside - if I were to play with the proportion of decoys in the database, (reducing the number to say 25% of the total database, rather than 50%), in your experience would this reduction in the search space be likely to have much of an effect on PSM / protein IDs? Thanks again, Pete On Tuesday, 5 June 2018 16:26:37 UTC-7, David Shteynberg wrote: > > Hello Pete, > > Use of decoys in the X!Tandem workflow is not strictly required but it is > helpful. X!Tandem is one of a handful of search engines for which there > exists a parametric mixture model for PSM-level validation. In my > experience, using the parametric models may not always give you the best > results in terms of validation, and the lack of decoys precludes the > ability to compute decoy-based error rates (FDR). In this case you have to > only the FDR as estimated by the PeptideProphet model, and you cannot > compare this FDR to decoy-based estimates. The best results are usually > obtained by using a database that includes some proportion of decoys. You > can then validate the search results using PeptideProphet's semi-supervised > semi-parametric models, which tend to estimate more accurate error rates. > However you decide to search your data, I also suggest you also include > iProphet in your validation analysis for peptide-level validation. > > Cheers, > -David > > On Tue, Jun 5, 2018 at 12:47 PM, <[email protected] <javascript:>> > wrote: > >> Please can someone advise - is it recommended to include decoys in my >> database when using X! for searches then peptide/protein prophet in TPP? >> >> >> As I understand, X! recommends not including decoys in searches (when >> used standalone) – however when I analyse X! identified peptides using >> peptide/protein prophet, I assume that I will need decoy matches to help >> pin down the correct distribution? >> >> >> Thanks >> >> Pete >> >> -- >> 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 post to this group, send email to [email protected] >> <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 [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.
