Cluster Analysis of Untargeted Metabolomic Experiments

Area of Science:

• Metabolomics
• Bioinformatics
• Systems Biology

Background:

• Untargeted metabolite profiling identifies unique metabolic phenotypes.
• These phenotypes can be associated with stress, disease, or environmental exposures.
• Unsupervised data analysis is crucial for quality control and biological insights.

Purpose of the Study:

• To demonstrate the utility of unsupervised data analysis in metabolomics.
• To show how to format untargeted mass spectrometry data for R.
• To explore the predictive power of data visualization techniques in biological samples.

Main Methods:

• Liquid chromatography-mass spectrometry (LC-MS) for untargeted metabolite profiling.
• Data formatting for import into the R statistical environment.
• Hierarchical clustering and principal component analysis (PCA) for data transformation and visualization.

Main Results:

• Unsupervised data analysis effectively identifies unique metabolic phenotypes.
• Visual representations of data using PCA and hierarchical clustering highlight biological sample variations.
• The methods allow for predictive insights into environmental stress and health.

Conclusions:

• Unsupervised data analysis is a powerful tool for quality control in metabolomics.
• Metabolite profiling combined with R-based analysis provides significant biological and health insights.
• This approach facilitates the understanding of cellular responses to various conditions.