Biology Reference
In-Depth Information
knowledge of the scientists. 62 As such, scientists
are to exercise prudence in determining the
adequacy of normalization, and a good guide
would be to observe tight clustering of the
pooled QCs among biological samples that are
highly scattered in the PCA score plot.
through utilization of statistical models to
derive optimal scaling factors for each sample
based on the entire dataset, such as normaliza-
tion by unit norm 58 or median 59 of intensity,
or total area normalization, 20 or maximum like-
lihood method. 60 These approaches, however,
suffer from the lack of an absolute concentration
reference for metabolites, poor consideration of
the nonself-averaging property of metabolites,
and alteration of covariance structure via con-
straining data to a speci
CONCLUSION AND
FUTURE OUTLOOK
c norm like total
signal. 61 Another approach is to normalize
against a single or multiple internal or external
standard compounds based on the retention
time regions or similarity in chemical properties
or metabolite classes. 47,55,61 Limitations in this
approach include challenges in the choice of
standards, assignment of the appropriate stan-
dards to normalize speci
GC/MS is an indispensible analytical tool
widely employed in metabonomics. The fast-
growing number of publications released each
year on GC/MS metabonomics is a good indi-
cator of the emerging popularity of its applica-
tion. The extension of its application from
small-scale studies to larger-scale (epidemiolog-
ical) studies to answer biological questions
clearly testi
c peaks in untargeted
metabonomics as retention time may not neces-
sarily be relevant to all chemical properties in
the matrix and normalization by a single stan-
dard is sensitive to its own obscuring varia-
tion. 52,61 Recently, Dunn et al. proposed an
approach termed quality control-based robust
LOESS signal correction (QC-RLSC) where
a low order nonlinear locally estimated smooth-
ing function is
es to the technical maturity of
GC/MS in metabonomics. 63 Although certain
challenges such as variation from derivatization
and metabolite identi
cation remain to be
resolved, it is heartening to observe researchers
exploring innovative strategies to overcome the
challenges imposed by analytical constraints
such as analytical drift. Such relentless efforts
would certainly elevate GC/MS-based metabo-
nomics to greater heights in the near future.
fitted to the pooled QC data with
respect to the order of injection to mitigate the
issue of signal intensity drift over time. 19 This
method requires the periodic analysis of pooled
QC samples together with study samples. A
correction curve for the whole analytical run is
interpolated and each detected peak in the total
data set is normalized to its respective peak in
QC sample using QC-RLSC. QC-RLSC has
been proposed to be able to facilitate data inte-
gration across the analytical block which is
important for large-scale studies. 19 It has been
commented that there is no such thing as the
correct or optimal normalization. 56 The ultimate
choice of the normalization method employed
depends on the nature of the variability in the
data, QC strategy used (internal standard or
external standard), and the preference and
References
1. Atkinson AJ, Colburn WA, DeGruttola VG, DeMets DL,
Downing GJ, Hoth DF, et al. Biomarkers and surrogate
endpoints: Preferred de
nitions and conceptual frame-
work. Clin Pharmacol Ther 2001; 69 (3):89 e 95.
2. Nicholson JK, Lindon JC, Holmes E.
:
understanding the metabolic responses of living
systems to pathophysiological stimuli via multivariate
statistical analysis of biological NMR spectroscopic
data. Xenobiotica 1999; 29 (11):1181 e 9.
3. Nicholson JK, Wilson ID. Understanding
Metabonomics
systems biology: metabonomics and the continuum of
metabolism. Nature Reviews Drug Discovery 2003; 2 (8):
668 e 76.
4. Pasikanti KK, Ho PC, Chan EC. Gas chromatography/
mass spectrometry in metabolic pro
global
ling of biological
Search WWH ::




Custom Search