Signed Distance Correlation (SiDCo): an online implementation of distance correlation and partial distance...

Area of Science:

• Bioinformatics
• Computational Biology
• Data Science

Background:

• Developing data-driven biological networks requires tools to measure complex relationships between metabolites.
• Existing tools primarily focus on linear correlations, neglecting non-linear dependencies.

Purpose of the Study:

• To present SIgned Distance COrrelation (SiDCo), a user-friendly GUI platform for calculating distance correlation in omics data.
• To enable the measurement of both linear and non-linear dependencies, even with varying sample sizes.
• To introduce a novel 'signed distance correlation' for enhanced metabolomic and lipidomic analyses.

Main Methods:

• Implementation of distance correlation for measuring linear and non-linear relationships.
• Development of a GUI platform (SiDCo) for omics data analysis.
• Integration of Pearson's correlation sign with distance correlation values to create signed distance correlation.
• Adaptation of Gaussian Graphical models for calculating partial distance correlation.

Main Results:

• SiDCo calculates distance correlation, capturing linear and non-linear associations between variables.
• The platform supports correlations between vectors of different lengths (e.g., varying sample sizes).
• Novel features include signed distance correlation and partial distance correlation, enhancing analytical capabilities.
• SiDCo offers one-to-one or one-to-all correlation analyses for comprehensive relationship mapping.

Conclusions:

• SiDCo provides an accessible and versatile platform for exploring complex relationships in omics data.
• The tool facilitates data-driven biological network development through advanced correlation measures.
• SiDCo is applicable to a wide range of datasets, with specific utility in metabolomics and lipidomics.