AutoXAI4Omics: an automated explainable AI tool for omics and tabular data.

Area of Science:

• Bioinformatics
• Computational Biology
• Artificial Intelligence in Biology

Background:

• Machine learning (ML) methods are increasingly vital for analyzing complex omics data.
• Researchers need user-friendly tools to leverage ML for biomarker identification and phenotype prediction without extensive coding expertise.

Purpose of the Study:

• To introduce AutoXAI4Omics, an automated, open-source explainable AI tool for omics and tabular data analysis.
• To enable researchers to perform classification and regression tasks, accelerating scientific discovery.

Main Methods:

• The tool automates ML pipeline processes, including feature selection, hyper-tuning, and model selection.
• It incorporates omic data-type-specific feature filtering.
• Explainability analysis provides insights into feature-target associations.

Main Results:

• AutoXAI4Omics automates complex ML decisions, saving researchers time.
• It facilitates the identification of novel, actionable insights by highlighting associations between omics features and predicted phenotypes.
• The tool supports both classification and regression tasks on omics and tabular numerical data.

Conclusions:

• AutoXAI4Omics empowers scientists to utilize sophisticated ML models for omics data analysis.
• The tool enhances biological interpretation and validation by providing explainable AI insights.
• It accelerates discovery by streamlining ML workflows and automating expert-level decisions.