Dave, Mike Hoopmann implemented a nice fix for searches using small bins. The sources current in Comet SourceForge trunk should be fully functional. One new parameter entry is titled "use_sparse_matrix" (see Comet version 2013.01 params in link below) which controls whether or not the sparse matrix data representation is used. For small bins, this not only addresses the memory use but those searches are also a little faster.
http://comet-ms.sourceforge.net/parameters/parameters_201301/ There's one other little feature I'd like to add before putting out another official release but the sparse matrix data representation to deal with this issue (memory use and small bin sizes) is the major feature in the next version. When the next release happens, I'll document memory use and compare small vs. large bins on high res data (among other analysis). - Jimmy On Mon, Apr 1, 2013 at 2:33 PM, Dave Trudgian < david.trudg...@utsouthwestern.edu> wrote: > Hi Philip, > > Sorry to drag up an old thread, but I wondered if you had info about the > effects on results of limiting Comet memory usage by using a large bin, vs > using the 64-bit binary and small bin? > > We are primarily working with QExactive data here. I was thinking of > adding Comet to our pipeline, and checking on the bin-size effect on > results - but cheekily wondered if you or anyone else have any feeling for > this? > > Cheers, > > Dave Trudgian > > > On Thursday, January 17, 2013 6:56:11 AM UTC-6, Philip Brownridge wrote: >> >> Hello Jimmy, thank you very much for your quick reply! You're completely >> correct about the fragment bin value, when I ran Comet (thank you for name >> guidance) using the LTQ settings it worked fine! When I ran it on the high >> resolution settings with task mamager on, I saw that I was getting the >> calloc error at about 2Gb of memory. I'm running Comet on a win7-64bit >> machine with 24Gb of memory and the program is called comet-win32, please >> could you tell me whether it would be possible to recompile Comet for win64 >> and perform larger searches? Sorry if this is a naive question, i'm very >> much a beginner when it comes to code. >> thank you again, >> Philip >> >> On Wednesday, 16 January 2013 19:55:58 UTC, Jimmy Eng wrote: >>> >>> Philip, >>> >>> I'm guessing you're specifying a small "fragment_bin_tol" value for the >>> high-res ms/ms spectra. This causes Comet (not all capitalized) to use a >>> ton of memory and you're just running out of memory. >>> >>> On the following UWPR SEQUEST page, there's a table correlating a set of >>> fragment_bin_tol settings vs. # input spectra vs. memory used: >>> https://proteomicsresource.**washington.edu/sequest_** >>> release/release_201201.php<https://proteomicsresource.washington.edu/sequest_release/release_201201.php> >>> These numbers apply to Comet as well. >>> >>> To address the problem, run smaller searches i.e. run Comet on a subset >>> of your input spectra. You can do this either with the "scan_range" >>> parameter in the params file or simply invoke the searches such as >>> comet yourfile.mzXML:1-3000 >>> comet yourfile.mzXML:3001-6000 >>> comet yourfile.mzXML:6001-9000 >>> >>> Your going to get multiple outputs if you do this which will need to be >>> handled. Splitting searches and managing results is something that some >>> script should do for you; some day I'll throw some simple windows and linux >>> example scripts on the Comet website for users. >>> >>> Hopefully we'll have a workaround to address the memory use for these >>> small fragment_bin_tol values in the semi-near future (thanks to Mike >>> Hoopmann here who has implemented sparse matrix support in the code which >>> I'm actively working with now). >>> >>> - Jimmy >>> >>> >>> On Wed, Jan 16, 2013 at 11:05 AM, Philip Brownridge < >>> philip.b...@gmail.com> wrote: >>> >>>> Hello all, please can I apologise if I have posted this in the wrong >>>> group, but I can't find a COMET group and there are other COMET postings >>>> here so I hope someone can help me! I'm trying to use some QExactive data >>>> with Comet and I keep getting either a calloc error message or a message >>>> saying there is no search to perform. I can get COMET searches to work on >>>> LTQ Orbitrap data but not QExactive. I have been using Proteowizard to >>>> convert the RAW files to mzML. I'm using the default params file with high >>>> resolution fragment ion parameters. >>>> Please if anybody is using COMET with QExactive data please could they >>>> let me know where i'm wrong! >>>> thanks in advance, >>>> Philip >>>> >>>> -- >>>> You received this message because you are subscribed to the Google >>>> Groups "spctools-discuss" group. >>>> To view this discussion on the web visit https://groups.google.com/d/** >>>> msg/spctools-discuss/-/**Jr5k5f4wcY0J<https://groups.google.com/d/msg/spctools-discuss/-/Jr5k5f4wcY0J> >>>> . >>>> To post to this group, send email to spctools...@googlegroups.com. >>>> To unsubscribe from this group, send email to spctools-discu...@** >>>> googlegroups.com. >>>> For more options, visit this group at http://groups.google.com/** >>>> group/spctools-discuss?hl=en<http://groups.google.com/group/spctools-discuss?hl=en> >>>> . >>>> >>> >>> -- > 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 http://groups.google.com/group/spctools-discuss?hl=en. > For more options, visit https://groups.google.com/groups/opt_out. > > > -- 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 http://groups.google.com/group/spctools-discuss?hl=en. For more options, visit https://groups.google.com/groups/opt_out.
