Review of Statistical Learning Methods in Integrated Omics Studies (An Integrated Information Science)

Area of Science:

• Modern biological science
• Computational biology
• Bioinformatics

Background:

• Integrated omics approaches are crucial for dissecting complex molecular systems.
• The emergence of statistical and machine learning methods has revolutionized integrated omics research.
• These methods integrate knowledge from biology, medicine, statistics, and artificial intelligence.

Purpose of the Study:

• To review statistical learning methods for integrated omics data.
• To streamline these methods within a statistical learning framework.
• To discuss the generalizability of these methods to other complex systems.

Main Methods:

• Review of 42 publications on exploratory and supervised learning methods for integrated omics.
• Discussion of extended principal component analysis, cluster analysis, network analysis, and regression methods.
• Inclusion of statistical techniques like penalization for sparsity and Bayesian approaches.

Main Results:

• Identified innovative and multidisciplinary statistical/machine learning methods for integrated omics.
• Highlighted the generalizability of these methods beyond biology to other complex systems.
• Emphasized the role of integrated omics within integrated information science.

Conclusions:

• Integrated omics provides a foundation for systematic learning and precision medicine.
• Statistical learning frameworks offer powerful tools for analyzing complex biological data.
• The reviewed methods and software packages support advanced omics data integration and analysis.