Biology Reference
In-Depth Information
3.5 Beyond Protein
Identifi cation:
Quantitation
and Extraction
of Biological Relevance
Quantitative proteomics can be divided into two categories: label-
free methods and the use of isotope labelling. Label-free pro-
teomics has several advantages over labelling techniques: all
organisms and sample types may be analyzed, the number of sam-
ples that can be compared is not limited, and data analysis is com-
paratively straightforward and does not require the use of specifi c
software. However, the volume of data accumulated in a global,
label-free proteomics study, such as in a GeLC-MS/MS experi-
ment, can be overwhelming. An important goal in functional pro-
teomics is to globally profi le changes in protein abundances in
biological systems and also provide a snapshot of the protein
expression state in response to biological perturbations. The even-
tual outcome of any quantitative proteomics study is to draw a
biological conclusion from the large volume of data acquired. This
can be systematically achieved by measuring the protein abundance
differences between proteins from two or more conditions, apply-
ing statistical tests for signifi cance, and visualizing results in a bio-
logical context (
see
ref.
28
for a recent review).
We analyze our data by spectral counting, specifi cally by calcu-
lating normalized spectral abundance factors (NSAFs) [
29
] for
each protein in a data set. This calculation takes into consideration
that the number of spectra identifi ed for a protein will be depen-
dent on the length of that protein. This calculation is incorporated
in a series of freely available R modules assembled in the form of
the Scrappy program [
27
]. Spectral counting using NSAFs has
been demonstrated to be an accurate method for calculating rela-
tive abundances of proteins between two or more samples [
30
-
32
]. In a quantitative study comparing spectral counting using
NSAFs with isobaric labelled peptides, we found that a much
greater volume of differentially expressed proteins were identifi ed
with the label-free method, although the two methods yielded
similar biological conclusions [
33
].
Extracting biological relevance from large-scale proteomics
data sets is a challenge in both label-free and labelled proteomic
studies. Numerous tools exist to functionally categorize protein
identifi cation data and map proteins to biological processes. In our
experience, we have found it useful to begin functional analysis by
accumulating Gene Ontology (GO) annotations [
34
] for proteins
identifi ed in the data set, followed by functional categorization
using Web Gene Ontology Annotation Plot (WEGO) [
35
]. WEGO
maps GO annotations to either functional pathways or processes
and calculates a p-value for category enrichment. Quantitative anal-
ysis of functional categories can be performed using PloGO, a freely
available, open-source tool [
24
]; rather than simply summing the
number of proteins in a category, the sum of NSAF values can be
used to estimate the relative abundance of proteins in a pathway or
a process. This information can be used to highlight areas of the
proteome that are most responsive to a stress or a specifi c treatment
and hence, that particular area may be studied in further detail.
