of proteomics including de novo analysis is the possibility to
obtain information about posttranslational modifications
(PTMs) present in analyzed proteins, something that is not
possible analyzing only the genome or transcriptome. For a
recent review of the subject, see Medzihradszky and Chalkley
[34]. PEAKS software contains the capability of running automatic de novo interpretation analysis of MS/MS spectra and
looks also for possible mutations and more than 400 PTMs.
Patternlab software also contains a specific module (Rapid
Novor) for running de novo analysis.
In addition, to provide a protein database, the different program (algorithms) for MS data analysis ask for (but some of them
already incorporate this information) a list of common contaminants that can be extracted from https://thegpm.org/cRAP and a
decoy database in order to calculate the False Discovery Rate
(FDR) for protein identifications. Once the MS data analysis is
finished, we would need to filter peptide/protein identification
results using an FDR threshold (usually an FDR < 1% at the
peptide spectrum matching (PSM) level). A protein identification
will be considered valid if based on reliable identification of at least
one unique peptide.
3.9 Mass
Spectrometry Data
Analysis: Protein
Quantification
Protein relative quantification (see Note 19) can be done on the
basis of both label-free and isobaric-labeling analyses, despite the
second ones are usually preferred, because they provide much
better precision, and also several important drawbacks have been
described for label-free approaches, the most important relates to
the missing value problem (see ref. 16). The need of multiple runs
decreases the precision of measurements due to well-known runto-run variation in MS analysis, affecting reproducibility, and
increasing substantially the number of missing values, which compromises the quantification of low abundant proteins [13, 14,
16]. In spite of this, label-free proteomic analysis is still a very
attractive approach for some specific experimental purposes due
to its simplicity when compared to sample multiplexing-based
approaches and the possibility of application to an unlimited number of samples (e.g., see refs. 35–37).
1. Quantification and normalization methods for label-free proteomic analysis
For quantitative label-free proteomic analysis where samples are analyzed individually, different strategies for extracting
quantitative data have been developed based on the area under
the curve (AUC; using signal intensity from precursor ion
spectra) or spectral count (counting MS/MS spectra that
match to different peptides, i.e., peptide spectrum matches
(PSMs)). Likewise, different normalization methods have
been proposed, for example, those that calculate correction
92
Angel P. Diz and Paula Sa ´ nchez-Marı ´n
obtain information about posttranslational modifications
(PTMs) present in analyzed proteins, something that is not
possible analyzing only the genome or transcriptome. For a
recent review of the subject, see Medzihradszky and Chalkley
[34]. PEAKS software contains the capability of running automatic de novo interpretation analysis of MS/MS spectra and
looks also for possible mutations and more than 400 PTMs.
Patternlab software also contains a specific module (Rapid
Novor) for running de novo analysis.
In addition, to provide a protein database, the different program (algorithms) for MS data analysis ask for (but some of them
already incorporate this information) a list of common contaminants that can be extracted from https://thegpm.org/cRAP and a
decoy database in order to calculate the False Discovery Rate
(FDR) for protein identifications. Once the MS data analysis is
finished, we would need to filter peptide/protein identification
results using an FDR threshold (usually an FDR < 1% at the
peptide spectrum matching (PSM) level). A protein identification
will be considered valid if based on reliable identification of at least
one unique peptide.
3.9 Mass
Spectrometry Data
Analysis: Protein
Quantification
Protein relative quantification (see Note 19) can be done on the
basis of both label-free and isobaric-labeling analyses, despite the
second ones are usually preferred, because they provide much
better precision, and also several important drawbacks have been
described for label-free approaches, the most important relates to
the missing value problem (see ref. 16). The need of multiple runs
decreases the precision of measurements due to well-known runto-run variation in MS analysis, affecting reproducibility, and
increasing substantially the number of missing values, which compromises the quantification of low abundant proteins [13, 14,
16]. In spite of this, label-free proteomic analysis is still a very
attractive approach for some specific experimental purposes due
to its simplicity when compared to sample multiplexing-based
approaches and the possibility of application to an unlimited number of samples (e.g., see refs. 35–37).
1. Quantification and normalization methods for label-free proteomic analysis
For quantitative label-free proteomic analysis where samples are analyzed individually, different strategies for extracting
quantitative data have been developed based on the area under
the curve (AUC; using signal intensity from precursor ion
spectra) or spectral count (counting MS/MS spectra that
match to different peptides, i.e., peptide spectrum matches
(PSMs)). Likewise, different normalization methods have
been proposed, for example, those that calculate correction
92
Angel P. Diz and Paula Sa ´ nchez-Marı ´n