Biology Reference
In-Depth Information
3. Protein Identification and Characterization
with MS Data
Protein identification plays a central role in the investigation of pro-
teomes, whenever using 2-DE gels or LC for sample separation followed
by MS analysis. Typically, the spots of a 2-DE gel are enzymatically
digested and the resulting peptide masses are measured, producing
an MS spectrum, also called “peptide mass fingerprinting” (PMF).
Peptides can also be isolated and fragmented within the mass spectrom-
eter, leading to MS/MS spectra, also called “Peptide Fragmentation
Fingerprinting” (PFFs).
3.1. PMF
In PMF analysis, peak lists extracted from an experimental spectrum are
compared with theoretical masses computed from protein sequences
stored in databases that were digested
in silico
using the same cleavage
specificity of the protease employed in the experiment. Practically, the
procedure counts matching experimental and theoretical peptide masses.
A protein is selected as a candidate when it contains a threshold-dependent
number of peptide hits. Other criteria, such as peptide coverage of the
sequence, mass error, etc., contribute to defining a score. The candidate
proteins are then sorted according to their scores. The top-ranked pro-
teins are considered as potential identifications of the spectrum.
The key step of the procedure lies in the scoring function, a higher
score indicating a higher likelihood that the corresponding protein is
the target one. Many factors are usually taken into account to produce
a robust score, such as dissimilarities in the peak positions due to inter-
nal or calibration errors or modified amino acids, expected peak inten-
sities, noise, contaminant or missing peaks, and so on. A variety of
different scoring schemes have been implemented in various algorithms
and integrated in current software. PeptideSearch
24
and PepFrag
25
use a
simple score based on the number of common masses between the
experimental and theoretical spectra. Pappin
et al
.
26
designed a scoring
function for the algorithm called MOWSE, which accounts for the
nonuniform distribution of protein and peptide molecular weights in
