-
Notifications
You must be signed in to change notification settings - Fork 7
Home
Welcome to the MSFragger wiki!
MSFragger is an ultrafast database search tool for peptide identifications in mass spectrometry-based proteomics. It differs from conventional search engines by computing similarity scores in a fragment-centric fashion using a theoretical fragment index of candidate peptides. The speed of MSFragger makes it particularly suitable for ‘open’ database searches, where the precursor mass tolerance is set to hundreds of Daltons, for the identification of modified peptides. MSFragger is implemented in the cross-platform Java programming language and is compatible with standard proteomics file formats such as MGF/mzXML/mzML/pepXML.
The processor requirements of MSFragger depends on the complexity of your search (and your patience to wait for search results). For an open search (500Da precursor mass window) using a tryptic digest of the human proteome, a single processor core can search roughly 40,000 MS/MS spectra in under an hour. MSFragger scales well with the number of processor cores and runtimes of under 2 minutes per file have been achieved using a 28-core workstation. A desktop workstation with a quad core processor is sufficient for most simple workflows.
MSFragger requires substantial amounts of memory due to its in-memory fragment index. While MSFragger can operate with less memory than needed to store the fragment index, it will cause index fragmentation where it breaks the search into multiple passes, searching each input file against a small segment of the index at a time (which greatly increases the runtime). For the human Uniprot protein database with reversed decoys, approximately 3700 MB of memory is needed to prevent index fragmentation. The actual size of the fragment index is substantially lower (MSFragger uses a very conservative estimate of the available free memory to avoid out of memory situations). Specifying common modifications may boost memory requirements to 6 GB. Semi-tryptic, non-enzymatic, and phospho searches may take tens of gigabytes to avoid fragmented searches. Limiting the range of peptide lengths can reduce the search space and reduce memory consumption in such cases. While fragment index fragmentation is undesirable, it may be unavoidable in certain instances.
We recommend at least 8GB of memory for workflows involving standard tryptic digestions.
MSFragger has been tested on Mac OS X, Windows 7, and a number of Linux distributions. Note that a 64-bit operating system is required to access more than 4GB of memory.
MSFragger is written using Java 1.8 and requires the Java 8 Runtime Environment. We recommend the Oracle Java 8 Runtime (download and installation instructions are available at www.java.com).