in one step. It is, therefore, often necessary to combine different
techniques to detect even a significant proportion of all metabolites
within a complex mixture [5]. Both gas chromatography-mass
spectrometry (GC-MS) and liquid chromatography-mass spectrometry (LC-MS) have been intensively used to profile broad
natural variance in the form of recombinant inbred lines (RILs),
introgression lines (ILs), and, more recently, genome-wide association mapping panels in order to enhance our understanding of the
regulation of plant primary and secondary metabolism (see reviews
[6, 7]).
In all metabolomics applications, it is important to understand
and control factors that contribute to sources of variation within
the data sets. The variability between samples can arise from multiple sources including natural biological variation itself and that
which occurs on sample collection and storage [8, 9]. In addition,
analytical variation caused by suboptimal performance of the apparatus used and instrument drift over time represent considerable
hurdles in large-scale metabolomics studies [10] Batch-to-batch
variation is the technical source of variation arising from the sum
of both manual and robotic samples handling [11]. The presence of
batch-to-batch variation makes it difficult to integrate data from
independent batches of samples. This issue is particularly problematic when dealing with large number of samples such as is the case
when analyzing structured plant populations.
To counter this, several normalization methods have been
developed and suggested to overcome these issues and to minimize
nonbiological variation [11–13]. For example, normalizations by a
single or multiple internal or external standard compounds based
on empirical rules, such as specific regions of retention time, have
been used [14]. Similarly, isotope-labeled internal standard
approaches were developed to monitor analytical error
[15]. While there is no single best practice to conduct metabolomic
studies, there are a number of pitfalls and known problems that
need to be carefully avoided. Detailed guidelines and practice and
normalization protocols have been published for this purpose [16–
18]. In this chapter, we describe a workflow for metabolomics
GWAS studies in Arabidopsis including growth and harvesting,
measurement of primary and secondary metabolites, association
mapping, and cross- or functional validation of the identified
associations.
Particularly the advent of genome-wide association studies, but
also the use of mapping populations has greatly facilitated: (1) The
association of genes to metabolic functions. (2) Both of structural
and regulatory genes such as transcription factors which have been
identified as controlling the accumulation of several specialized
metabolite. (3) GWAS has become a key component of investigations into the interaction of metabolites with phenotypic traits such
as growth [19] and the rates of photosynthesis and respiration
394
Si Wu et al.
techniques to detect even a significant proportion of all metabolites
within a complex mixture [5]. Both gas chromatography-mass
spectrometry (GC-MS) and liquid chromatography-mass spectrometry (LC-MS) have been intensively used to profile broad
natural variance in the form of recombinant inbred lines (RILs),
introgression lines (ILs), and, more recently, genome-wide association mapping panels in order to enhance our understanding of the
regulation of plant primary and secondary metabolism (see reviews
[6, 7]).
In all metabolomics applications, it is important to understand
and control factors that contribute to sources of variation within
the data sets. The variability between samples can arise from multiple sources including natural biological variation itself and that
which occurs on sample collection and storage [8, 9]. In addition,
analytical variation caused by suboptimal performance of the apparatus used and instrument drift over time represent considerable
hurdles in large-scale metabolomics studies [10] Batch-to-batch
variation is the technical source of variation arising from the sum
of both manual and robotic samples handling [11]. The presence of
batch-to-batch variation makes it difficult to integrate data from
independent batches of samples. This issue is particularly problematic when dealing with large number of samples such as is the case
when analyzing structured plant populations.
To counter this, several normalization methods have been
developed and suggested to overcome these issues and to minimize
nonbiological variation [11–13]. For example, normalizations by a
single or multiple internal or external standard compounds based
on empirical rules, such as specific regions of retention time, have
been used [14]. Similarly, isotope-labeled internal standard
approaches were developed to monitor analytical error
[15]. While there is no single best practice to conduct metabolomic
studies, there are a number of pitfalls and known problems that
need to be carefully avoided. Detailed guidelines and practice and
normalization protocols have been published for this purpose [16–
18]. In this chapter, we describe a workflow for metabolomics
GWAS studies in Arabidopsis including growth and harvesting,
measurement of primary and secondary metabolites, association
mapping, and cross- or functional validation of the identified
associations.
Particularly the advent of genome-wide association studies, but
also the use of mapping populations has greatly facilitated: (1) The
association of genes to metabolic functions. (2) Both of structural
and regulatory genes such as transcription factors which have been
identified as controlling the accumulation of several specialized
metabolite. (3) GWAS has become a key component of investigations into the interaction of metabolites with phenotypic traits such
as growth [19] and the rates of photosynthesis and respiration
394
Si Wu et al.