Multi-omic integration by machine learning (MIMaL).

Area of Science:

• Systems Biology
• Computational Biology
• Molecular Biology

Background:

• Cells dynamically regulate gene expression in response to environmental cues.
• Multi-omics approaches integrate diverse molecular data (RNA, proteins, metabolites) for a holistic view of biological systems.
• Integrating proteome and metabolome data presents challenges in knowledge extraction.

Purpose of the Study:

• To develop and apply a novel method for integrating multi-omics data, specifically proteome and metabolome data.
• To identify novel connections between proteins and metabolites, revealing protein control over metabolic pathways.
• To predict and experimentally validate gene functions using multi-omic data integration.

Main Methods:

• Utilized machine learning and model interpretation to analyze multi-omics datasets.
• Developed the Multi-omic Integration and Machine Learning (MIMaL) approach.
• Mapped identified protein regulators onto existing genetic and metabolic networks.

Main Results:

• Discovered novel protein regulators (ProC) controlling metabolites, including citrate.
• Identified previously unknown protein regulators and validated their roles.
• Successfully predicted and experimentally confirmed functions for five genes, including two uncharacterized genes (YJR120W, YDL157C) involved in mitochondrial translation.

Conclusions:

• The MIMaL approach effectively integrates proteomic and metabolomic data to reveal novel biological insights.
• This method facilitates the discovery of novel gene functions and regulatory mechanisms.
• A publicly available website and code enable broader application of MIMaL for multi-omics data analysis.