function of different sources of variation including variation
introduced by differences in sample collection. Large-scale
experiments with vast sample size and genotype number (e.g.,
ILs, RILs or GWAS) have the added complication that some
individuals may differ somewhat in their developmental age.
However, having a large number of biological replicates represents an essential means to minimize metabolite variation during sample preparation even if it cannot account for all sources
of this variation.
6. In case the experiment is too large to allow harvesting in a
single and relatively short time, it is essential to harvest control
samples from each temporally separate harvest. Further, plant
metabolomics experiment are generally performed at the organ
levels, and recommend to harvest pooled samples (several fruit
or leafs) per biological replicate. In addition, the age or developmental stage should be carefully considered according to
standardized growth condition and phenology descriptors.
7. All samples for a given experiment should follow exactly the
same procedure before grinding, during extraction, and analyzing. Sample grinding is usually required to optimize solvent
extraction and additionally aids in homogenizing the sample
material [37]. Many extraction protocols are available for plant
metabolomics and have been discussed in detail before (for
example, see Shimizu et al. [29] for LC-MS, and [38] for
GC-MS.). However, there are some important points at
which these protocols should be adapted when handling the
large number of plant samples required for QTL analysis. First,
quality control (QC) (see Note 7) is necessary throughout the
entire sample preparation process, from the greenhouse to the
sample storage location and through distribution to chemical
analysis and data normalization to reduce the analytical errors.
8. The quality control (QC) samples should represent, qualitatively and quantitatively, the entire samples included in the
study and should provide an average of all of the metabolomes
measured in the study. This can be done by pooling aliquots of
individual samples, either all or a subset representative for the
study. The QC samples should be evenly distributed over all the
batches and are extracted, derivatized, and analyzed at the same
time as the individual study samples. The data from the QC
samples is used to monitor drift, separate high- and low-quality
data, equilibrate the analytical platform, correct for drift in the
signal, and allow the integration of multiple analytical experiments. The data analysis technique such as principal component analysis can be used to quickly assess the reproducibility of
the QC samples in an analytical run. The QC samples are used
to determine the variance of a metabolite feature.
404
Si Wu et al.
introduced by differences in sample collection. Large-scale
experiments with vast sample size and genotype number (e.g.,
ILs, RILs or GWAS) have the added complication that some
individuals may differ somewhat in their developmental age.
However, having a large number of biological replicates represents an essential means to minimize metabolite variation during sample preparation even if it cannot account for all sources
of this variation.
6. In case the experiment is too large to allow harvesting in a
single and relatively short time, it is essential to harvest control
samples from each temporally separate harvest. Further, plant
metabolomics experiment are generally performed at the organ
levels, and recommend to harvest pooled samples (several fruit
or leafs) per biological replicate. In addition, the age or developmental stage should be carefully considered according to
standardized growth condition and phenology descriptors.
7. All samples for a given experiment should follow exactly the
same procedure before grinding, during extraction, and analyzing. Sample grinding is usually required to optimize solvent
extraction and additionally aids in homogenizing the sample
material [37]. Many extraction protocols are available for plant
metabolomics and have been discussed in detail before (for
example, see Shimizu et al. [29] for LC-MS, and [38] for
GC-MS.). However, there are some important points at
which these protocols should be adapted when handling the
large number of plant samples required for QTL analysis. First,
quality control (QC) (see Note 7) is necessary throughout the
entire sample preparation process, from the greenhouse to the
sample storage location and through distribution to chemical
analysis and data normalization to reduce the analytical errors.
8. The quality control (QC) samples should represent, qualitatively and quantitatively, the entire samples included in the
study and should provide an average of all of the metabolomes
measured in the study. This can be done by pooling aliquots of
individual samples, either all or a subset representative for the
study. The QC samples should be evenly distributed over all the
batches and are extracted, derivatized, and analyzed at the same
time as the individual study samples. The data from the QC
samples is used to monitor drift, separate high- and low-quality
data, equilibrate the analytical platform, correct for drift in the
signal, and allow the integration of multiple analytical experiments. The data analysis technique such as principal component analysis can be used to quickly assess the reproducibility of
the QC samples in an analytical run. The QC samples are used
to determine the variance of a metabolite feature.
404
Si Wu et al.