Multi-omic modelling of inflammatory bowel disease with regularized canonical correlation analysis

Area of Science:

• Genomics
• Microbiome Research
• Computational Biology

Background:

• Personalized medicine relies on understanding patient phenotype and predicting outcomes.
• Identifying variable interactions is challenging without prior hypotheses.
• Complex diseases like Crohn's disease involve intricate multi-omic relationships.

Purpose of the Study:

• To develop a method for identifying relationships between microbiome, host transcriptome, and clinical variables.
• To apply this method to Crohn's disease data.
• To determine the importance of model selection in multi-omic analysis.

Main Methods:

• Utilized a novel method based on canonical correlation analysis.
• Tested and optimized parameters using glioma and inflammatory bowel disease datasets.
• Applied the selected model to a Crohn's disease cohort with longitudinal sampling.

Main Results:

• Identified interactions between transcriptome, gut microbiome, and clinical variables.
• Demonstrated that model selection significantly impacts the identification of these relationships.
• Showed that incorporating clinical variables enhanced the model's explanatory power compared to co-inertia analysis.

Conclusions:

• The presented methodology offers a framework for analyzing multi-omic and clinical data interactions.
• Specific microbial and genetic factors associated with Crohn's disease were identified.
• Emphasized the critical role of appropriate model selection in multi-omics research.