CODARFE: Unlocking the prediction of continuous environmental variables based on microbiome.

Area of Science:

• Microbiology
• Bioinformatics
• Environmental Science

Background:

• Limited tools exist for analyzing microbiome data and correlating taxonomic composition with environmental factors.
• Current methods fail to predict environmental factors in new samples, creating a significant knowledge gap.
• There is a need for advanced solutions to understand microbiome dynamics and bridge the prediction gap.

Purpose of the Study:

• Introduce CODARFE, a novel computational tool for microbiome analysis.
• Enable sparse compositional microbiome predictor selection.
• Facilitate the prediction of continuous environmental factors from microbiome data.

Main Methods:

• Evaluated CODARFE against four state-of-the-art tools.
• Tested predictor selection accuracy across 24 diverse databases.
• Assessed the tool's performance in predicting environmental factors using cross-study validation.

Main Results:

• CODARFE outperformed existing tools in predictor selection in 21 of 24 databases.
• Identified significantly more bacteria linked to environmental factors in human data (at least 7% increase).
• Achieved an 11% mean absolute percentage error in cross-study environmental factor prediction.

Conclusions:

• CODARFE demonstrates robustness and broad applicability across various scientific fields and experimental conditions.
• The tool's predictive capabilities enable novel insights in unexplored research contexts.
• CODARFE provides a versatile solution for researchers seeking to deepen their understanding of microbiome-environment interactions.