Quantification of Proteins by Label-Free LC-MS/MS - LC-MS/MS in Proteomics: Methods and Applications - page 224

14 predetermined models of specific, physicochemical attributes
(such as retention time and fragmentation pattern). All tentative
peptides are collapsed into their parent proteins utilizing only the
highest scoring peptides that contribute to the total protein score.
Once a protein has been identified, all top-ranked precursor ions
and their corresponding product ions are removed from all other
tentatively identified proteins. The remaining unidentified pep-
tides - and tentatively identified proteins - are then re-ranked
and re-scored, and the process is repeated until a 4% false-positive
rate is reached. The false discovery rate is determined by the num-
ber of random or reverse identifications identified (false-positive
rate - FPR) divided by the number of correct identifications (true-
positive rate - TPR), expressed as a percentage. Therefore the
false discovery rate (FDR) is given by FDR

=

FPR/TPRx100.
All protein identifications were based on at least two peptides.
Since this search algorithm is not probability based there is no
need for a crude cutoff for selection of individual MS/MS spectra.

3.3.DataAnalysis

A typical quantitative analysis includes dozens of samples and
thousands of peptides. Thus the data analysis scheme should be
automated and unbiased. This can be done using statistical soft-
ware packages such as the free software R ( www.r-project.org ).
The output of the analysis will highlight the most significantly
changing proteins and peptides depending on significance thresh-
olds. There are five major steps to perform: alignment, nor-
malization, filtering, annotation, and combining. The very last
step is statistical analysis. Since the choice of statistical methods
can vary based on the experimental design and purpose, it is
excluded from this protocol. It is, however, recommended that
for a comparison of two treatment groups, univariate statistics
such as two-tailed, unpaired Student's t test is used after a loga-
rithmic transformation to approximate a normal distribution. For
multivariate statistics a partial-least square discriminant analysis
(PLS-DA) can be used to find significantly changed peptides or
proteins.

A key step in label-free LC-MS-based quantitation is the time
alignment and annotation of data. Since all samples are ana-
lyzed sequentially and separately, the data must be combined
and summarized. There are several commercially available soft-
ware packages that can perform this alignment. These include
Elucidator © (Rosetta Biosoftware), Progenesis LC-MS © (Nonlin-
ear Dynamics), and Proteinlynx Global Server © (Waters, Milford,
MA). The result of the time alignment is a two-dimensional table
that includes all detected peptides, their average mass, average
retention time, and detected intensity in all replicates of all sam-
ples. This table is the basis for all consequent data analysis steps
that follow ( see Fig. 13.2 ) . This table can then be annotated using

3.3.1.TimeAlignment