Comments on LC-MS/MS Presets
Default MS-BLAST Search settings (you see them once opening
the front page) have been optimized for sequence-similarity identifications with
relatively small and accurate peptide queries, which are typically
produced by accurate interpretation of good quality MS/MS spectra from
"hand-picked" precursors. These settings maximize the sensitivity
of sequence-similarity identifications and strictly follow MS-BLAST scoring
scheme (Habermann et al., 2004). Please note, however, that the
scheme relies on the threshold scores determined in computational experiments
using model, small size queries. This, however, could hardly be expanded to ca
100-fold larger queries produced by automated interpretation of hundreds of
MS/MS spectra from data-dependent LC-MS/MS (Waridel et al., 2007) and
therefore MS-BLAST settings required some practical adjustments. This is
achieved by activating LC-MS/MS Presets box, which changes
several important MS-BLAST settings:
- S and S2 are score thresholds for,
respectively, the highest and all other HSPs reported for the database search
hit. Specifying the threshold at reasonably high values prevents MS-BLAST from
reporting many weakly matching HSPs that plug the scoring scheme and increase
the false positive rate. However, if necessary, weaker alignments might still be
included by lowering the corresponding thresholds.
- B, V, hspmax are,
respectively, the number of allowed alignments, descriptions and HSPs. By
default, they have been set at the arbitrary value of 1000, which, in our
experience, should suffice for processing queries compiled by the automated
interpretation of 500-600 MS/MS spectra. Note that unnecessary high settings
slows down the searches considerably and only increases the number of reported
non-confident alignments. Use higher settings only if MS-BLAST produces a
warning message if the limit for B, V and
hspmax was exceeded.
- Filtering of low complexity sequences has been engaged by setting it to
default. This filter, effectively, eliminates low complexity
sequence stretches that are common in human and sheep keratin peptides -
ubiquitous contaminants commonly encountered in the analysis of in-gel digests.
The filter does not delete these peptide, but substitutes the corresponding
regions by zero-scoring X symbols - they could be recognized in the input
query reported at the top of MS BLAST output page. Albeit low complexity
filtering reduces the number of keratin-related hits, it could accidentally
eliminate bona fide proteins. Hence, repeating the search with the low
complexity filter turned off, it is recommended for the most complete
characterization of the analyzed sample - note, however, that this search
might require setting higher B, V and
hspmax values.
References:
- Habermann, B., Oegema, J., Sunyaev, S., and Shevchenko, A. (2004). The power
and the limitations of cross-species protein identification by
mass spectrometry-driven sequence similarity searches. Mol Cell
Proteomics 3, 238-249.
- Waridel, P., Frank, A., Thomas, H., Surendranath, V., Sunyaev, S., Pevzner,
P., and Shevchenko, A. (2007). Sequence similarity-driven proteomics in
organisms with unknown genomes by LC-MS/MS and automated de novo
sequencing. Proteomics 7, in press.
|