Comet release 2014.01
Documentation for parameters for release 2014.01
can be found here.
release 2014.01 rev. 1 (2014.01.1), release date 2014/06/03
- Known issue: when using "precursor_tolerance_type = 1"
to specify the tolerance is applied on the precursor m/z peak,
Comet has been errantly calculating the mass tolerance on the
deconvoluted peptide mass and then additionally scaling this
tolerance by the charge state. This inflates the precursor
tolerance by N-fold where N is the precursor charge state (so a
10 ppm tolerance ends up being 20 ppm for 2+ precursors and
30 ppm for 3+ precursors). The fix is to always specify
"precursor_tolerance_type = 0".
This parameter will be deprecated in the next release.
- Known issue: using custom amino acids (B, J, U, X, Z) fails.
- Known issue: high-res ms/ms searches (using small
"fragment_bin_tol"
values) are ~2x slower than 2014.01.0 due to an unnecessary array
initialization that was added in this maintenance release.
- Known issue: Using -N<name> command line option adds full
path to "spectrum" attribute in Windows.
- Known issue: Using -N<name> command line option, the pep.xml
"base_name" attribute has full path to the output file instead of
to the input file.
- Re-use temporary arrays during spectral preprocessing for better memory
management running under Windows, implemented by M. Hoopmann.
- New parameter
"override_charge"
instructs Comet to override the listed precursor charge state in the input
file with that specified by the
"charge_range"
parameter, implemented by D. Shteynberg.
- Update MSToolkit to version r68 to support non-consecutive mzXML scans.
- Bug fix: synchronize normal and sparse matrix xcorr scores when decoys are
generated to fill in the xcorr distribution for the expectation score
calculation, reported by D. Tabb.
- Bug fix: buffer overflow issue on some linux distributions when retrieving
host name, reported by K. Tamura.
- Bug fix: close spawned threads, reported by M. Riffle.
- Bug fix: corrected "sample_enzyme" reporting in pep.xml output, reported
by L. Mendoza.
- Bug fix: remove extra tab in txt output, reported by A. Kertesz-Farkas.
- Bug fix: output file name gets mangled when .out files are used and the
input is specified using a full path, reported by C. Hoogland.
release 2014.01 rev. 0 (2014.01.0), release date 2014/03/25
- Known issue: the sample_enzyme information in pep.xml output is completely broken.
- Known issue: there is a memory leak in the program. Avoid invoking
Comet with a ton of input files on the command line.
- Known issue: report of inconsistent results using
"sparse_matrix = 0" and
"sparse_matrix = 1"
searches. I believe this is due to decoy xcorr score generation, affecting
E-values, in cases where there is a low number of candidate peptide matches.
- Known issue: it's been reported that results are inconsistent
depending on the number of threads used in a search (Windows). Some numbers
deviate in the 3rd/4th decimal point.
- Add "output_suffix"
parameter option which appends a suffix to the
base name of pep.xml, pin.xml, txt, and sqt output files.
- Add Native ID attribute to pep.xml output when searching mzML files.
This entails an update to the MSToolkit
file parsing library which was made by M. Hoopman.
- Report the name of the file that's being searched in the runtime output.
- Avoid creating empty/stub output files in the case where an input
file has been moved away during a search. These first four features
were based on requests by D. Tabb.
- Change how E-values are calculated when
"decoy_search = 2"
is applied. This setting performs a separate target and decoy search as if
two different databases were searched separately resulting in two sets of
search results. In the past, the E-value calculations for the target hits
were calculated on the underlying target xcorr score distribution only. And
the decoy E-value calculations were based only on the decoy xcorr score
distribution. Now, the target and decoy score hits contribute to one xcorr
score distribution that is used to calculate the target and decoy E-values
per suggestion by the Noble lab.
- Sort results by scan number; previously results were reported in order of
increasing peptide mass. With results sorted by scan number, this allows
more efficient analysis using quantification tools that cache spectral
data in memory. Requested by D. Shteynberg.
- Bug fix: Change in spectral processing to handle edge case where the
last peak in some spectra are excluded from the analysis.
- Bug fix: Correct a bug that required order of two loops to be swapped.
Before this fix, a small number of fragment ions possibly do not get
considered for particular peptide/spectrum/charge combinations.
Thanks to A. Kertesz-Farkas for identifying these last two issues.
- For Windows binary only: revert previous memory changes (using VirtualAlloc
and VirtualLock functions) as there were negative consequences associated
with unexpected out of memory errors.
- For Crux compiled version only: modify text output, primarily changing column
order using changes submitted by S. McIlWain.
- For Crux compiled version only: report all peptide hits, including those with
negative xcorr scores.
- As the "output_suffix"
is the only new parameter entry which most researchers likely won't use,
this version of Comet will also run with comet.params files from
version 2013.02.
Go download from the download page.