9. In GC-MS, several tools, software, and data bases have been
established and used for this purpose (for more details see [26–
28]. For details on data processing of LC-MS data see [28]. For
both methods, manual checking of the peaks is strongly recommended. REFINER MS® 10.0 (GeneData, http://www.
genedata.com) can be used to analyze Chromatograms from
the UPLC–FT/MS. From the raw files, the molecular masses,
retention time, and associated peak intensities for each sample
are extracted. The chromatogram alignments are performed
using a pairwise alignment-based tree using m/z windows of
five points and RT windows of five scans within a sliding frame
of 200 scans. Resulting data matrices with peak ID, retention
time, and peak intensities in each sample are generated.
10. The goal of metabolomics as a phenotyping platform depends
on its ability to detect biologically related metabolite changes
in complex biological samples. As with any high-throughput
technology, systematic biases are often observed in LC-MS and
GC-MS metabolomics data [16, 28, 39]. As the number of
samples in the data set increases, there is a corresponding timedependent variation in the metabolite data. Removing
platform-specific sources of variability such as systematic errors
is hence one of the top priorities in metabolomics data preprocessing. However, metabolite diversity leads to different
responses to variations in given experimental conditions,
making normalization a very demanding task [40]. For the
effective elimination of different sources of analytical variation,
preprocessing steps should follow a specific sequence. Here the
Quality Control (QC) samples are of prime importance, and
these are best prepared by pooling equal volumes of material
from all of the biological samples to be analyzed. Alternatively,
a chemically defined mixture of authenticated reference compounds [41] that mimics the metabolic composition of the
investigated biological material can be employed [28]. Both
the synthetic mixtures and biological QC samples should then
be subjected to the same sample extraction, instrumental analyses (ideally distributed across the analytical run), and data
processing, thus providing quality checks for technical and
analytical error, and quantitative calibration to eliminate
batch effects for the final processed data. This normalization
is a crucial step for minimizing the batch-to-batch data variability across extended periods.
11. GWAS enables the analysis of associations between hundreds of
thousands of single-nucleotide polymorphisms (SNPs) and
specific traits. First the DNA obtained from hundreds to
thousands of natural genetic accessions, and the accessions
are genotypes and the genetic variations (like SNPs) are determined across the accessions. The principle based on If certain
Metabolomic Analysis of Natural Variation in Arabidopsis
405
established and used for this purpose (for more details see [26–
28]. For details on data processing of LC-MS data see [28]. For
both methods, manual checking of the peaks is strongly recommended. REFINER MS® 10.0 (GeneData, http://www.
genedata.com) can be used to analyze Chromatograms from
the UPLC–FT/MS. From the raw files, the molecular masses,
retention time, and associated peak intensities for each sample
are extracted. The chromatogram alignments are performed
using a pairwise alignment-based tree using m/z windows of
five points and RT windows of five scans within a sliding frame
of 200 scans. Resulting data matrices with peak ID, retention
time, and peak intensities in each sample are generated.
10. The goal of metabolomics as a phenotyping platform depends
on its ability to detect biologically related metabolite changes
in complex biological samples. As with any high-throughput
technology, systematic biases are often observed in LC-MS and
GC-MS metabolomics data [16, 28, 39]. As the number of
samples in the data set increases, there is a corresponding timedependent variation in the metabolite data. Removing
platform-specific sources of variability such as systematic errors
is hence one of the top priorities in metabolomics data preprocessing. However, metabolite diversity leads to different
responses to variations in given experimental conditions,
making normalization a very demanding task [40]. For the
effective elimination of different sources of analytical variation,
preprocessing steps should follow a specific sequence. Here the
Quality Control (QC) samples are of prime importance, and
these are best prepared by pooling equal volumes of material
from all of the biological samples to be analyzed. Alternatively,
a chemically defined mixture of authenticated reference compounds [41] that mimics the metabolic composition of the
investigated biological material can be employed [28]. Both
the synthetic mixtures and biological QC samples should then
be subjected to the same sample extraction, instrumental analyses (ideally distributed across the analytical run), and data
processing, thus providing quality checks for technical and
analytical error, and quantitative calibration to eliminate
batch effects for the final processed data. This normalization
is a crucial step for minimizing the batch-to-batch data variability across extended periods.
11. GWAS enables the analysis of associations between hundreds of
thousands of single-nucleotide polymorphisms (SNPs) and
specific traits. First the DNA obtained from hundreds to
thousands of natural genetic accessions, and the accessions
are genotypes and the genetic variations (like SNPs) are determined across the accessions. The principle based on If certain
Metabolomic Analysis of Natural Variation in Arabidopsis
405