Strategy for improved characterization of human metabolic phenotypes using a COmbined Multi-block Principal...

Area of Science:

• Metabolomics
• Systems Biology
• Bioinformatics

Background:

• Large-scale population omics data offer insights into gene-environment interactions and disease.
• Current dimension reduction techniques struggle with detailed information extraction from complex omics datasets.

Purpose of the Study:

• To introduce a novel interactive software pipeline for exploratory analysis of population-based nuclear magnetic resonance (NMR) spectral data.
• To enable efficient screening of population datasets for identifying molecular features linked to health and lifestyle factors.

Main Methods:

• Utilized COmbined Multi-block Principal components Analysis with Statistical Spectroscopy (COMPASS) within the R-library hastaLaVista.
• Applied Principal Component Analysis (PCA) to sequential spectral regions (blocks) for sub-population identification.
• Employed Statistical TOtal Correlation SpectroscopY (STOCKS) for molecular identification of differentiating signals.

Main Results:

• Developed a semi-automated pipeline for granular analysis of NMR spectral data.
• Successfully identified sub-populations and key molecular features related to drug intake, latent diseases, and diet.
• Provided population statistics for underlying feature distributions.

Conclusions:

• COMPASS offers an efficient and semi-automated approach for screening large population-based NMR datasets.
• The pipeline facilitates detailed exploratory analyses, improving understanding of gene-environment-disease interactions.