mosGraphGen: a novel tool to generate multi-omics signaling graphs to facilitate integrative and interpretable graph...

Area of Science:

• Computational Biology
• Bioinformatics
• Artificial Intelligence in Medicine

Background:

• Multi-omics data (genomics, epigenomics, transcriptomics, proteomics) offer a comprehensive view of cellular signaling but are challenging to integrate and interpret.
• Graph AI models are well-suited for multi-omics data integration due to their ability to represent biological networks.
• Converting multi-omics data into biologically meaningful graphs for Graph AI is a significant hurdle.

Purpose of the Study:

• To develop a novel tool, mosGraphGen, for automated generation of Multi-omics Signaling graphs (mos-graphs).
• To facilitate the direct application and evaluation of Graph AI models on multi-omics data.
• To streamline the process of mining key disease targets and signaling pathways from complex biological datasets.

Main Methods:

• mosGraphGen maps multi-omics data from individual samples onto a pre-defined, biologically meaningful signaling network.
• Data normalization is performed by aggregating measurements and aligning them to a reference genome.
• The tool generates sample-specific mos-graphs ready for input into Graph AI models.

Main Results:

• mosGraphGen successfully generated mos-graphs from TCGA and Alzheimer's disease (AD) multi-omics datasets.
• The generated mos-graphs enable direct use by AI model developers for analysis.
• The open-source availability of mosGraphGen promotes wider adoption and research.

Conclusions:

• mosGraphGen effectively addresses the challenge of preparing multi-omics data for Graph AI analysis.
• The tool standardizes the generation of signaling graphs, enhancing the interpretability of multi-omics data.
• mosGraphGen is a valuable resource for researchers aiming to leverage AI for biological discovery from complex omics data.