Sunday, April 7, 2013

Early impressions on MSAmanda vs. Sequest and Mascot

As mentioned in the title, this is an early impression.  Since we're currently drafting this up for publication, I choose not to go too far into the details.  Believe me when I say that this is extremely representative of what I am currently seeing in PD 1.4 when I use all 3 available search engines on phospho data from HR MS/MS spectra in conjunction with PhosphoRS.  I am seeing, of course, a lot of overlap but my early analyses have suggested that adding MSAmanda is comparable to the addition of a Mascot server to my processing.  I don't know how much Mascot costs, but I do know how much MSAmanda costs, $0 and about 3 minutes of your time to download and install it.  One of the best parts?  It works almost exactly like the Sequest node.  All of your modifications, reagents, and databases are all right there and you just pull them into the node the same way you do the Sequest one.

I'll save the sample processing details for the manuscript, but I will say that this is actually not a very fair comparison for MSAmanda.  This is because of FDR.  On Sequest HT and Mascot, Percolator gave me a significant boost, giving me 20-40% more peptides than using a simple 1% FDR.  For some reason I haven't investigated yet, searches of this data set on MSAmanda with Percolator do not complete. So this is MSAmanda with a 1% FDR and both of the other data sets boosted with Percolator cutoffs.
 If you use 1% FDR on all three of them:  MSAmanda wins by far, finding dramatically more peptides than either Mascot or Sequest.  I am using the data above because I am more concerned with total peptides right now than dumping a lot of time into troubleshooting FDRs.  I am a pathway-centric biologist, after all.

BTW, yes these numbers are low.  When we see phosphoproteomics papers we like to see thousands of sites, there is a good reason for this, but I can't go into it without revealing what I'm working on.

Anyway!  Download MSAmanda and instructions for installing it in Proteome Discoverer here.

UPDATE:  4/16/13:   Scratch what I've said about MSAmanda and Percolator.  It works just fine with Percolator.  You simply need to make sure that you aren't ordering the same modification twice.  I set Carbamidomethylation of cysteine as both a static and dynamic mod.  That would error out just about everything.
Also, if you download MSAmanda now, it installs with no problems on Win 7 32-bit PCs.  Again, it may have been my mistake.  At the end of my incredibly long days I get to work on my own projects.  Sometimes I'm really tired and sleepy by then....

5 comments:

  1. Hi Ben,

    I do observed that MS Amanda identifies more than Sequest or Mascot. How much error window are you using? MS Amanda works better for high accuracy data. For. e.g. it gave me more i.d. using 10ppm and 0.02Da (MS and MS/MS error) than 20ppm and 0.1Da..

    -
    Santosh

    ReplyDelete
  2. Santosh,
    Thats really interesting. I experimented with some different thresholds but didn't see much of a boost when I cut the mass range down. Since I've only been doing phoospho studies I wonder if that has something to do with it. I have yet to try it for processing a normal (no PTMs) data set.
    -Ben

    ReplyDelete
    Replies
    1. Dear Ben
      did you also try "MS2 Spectrum Processor, v. 0.5"?
      it appears to improve hcd outcomes
      andrea

      Delete
  3. Dear Ben,
    please download newest version of MS Amanda. Should boost your low scoring phospho peptides again.
    If you have specific question there is now a google MS Amanda site available.
    best
    Karl

    ReplyDelete
  4. Hello everyone. Is there a 64-bit version available?

    Best regards,

    Suzanne

    ReplyDelete