Probabilistic classification of gene-by-treatment interactions on molecular count phenotypes

Area of Science:

• Genomics
• Systems Biology
• Biomedical Research

Background:

• Genetic variation influences responses to treatments (G×T) and environmental stimuli (G×E).
• Response molecular QTL mapping identifies G×T signals for molecular phenotypes like gene expression.
• Current methods lack precision in distinguishing types of genetic effects under treatment.

Purpose of the Study:

• To develop a method for classifying response molecular QTLs into subclasses with distinct genetic interpretations.
• To provide a more nuanced understanding of genotype-by-treatment interactions.
• To improve the interpretation of molecular QTL studies in biomedicine.

Main Methods:

• Utilized Bayesian model selection to classify G×T interactions.
• Assigned posterior probabilities to different types of G×T interactions for feature-SNP pairs.
• Compared linear and nonlinear regression models for log-scale counts, incorporating genotype-phenotype relationships.

Main Results:

• The developed method offers an intuitive and well-powered framework for reporting and interpreting G×T interactions.
• Demonstrated the method's efficacy through simulations and application to existing datasets.
• Successfully classified response molecular QTLs into meaningful subclasses.

Conclusions:

• The new classification method enhances the interpretation of G×T interactions in molecular QTL studies.
• Provides a valuable tool for researchers investigating genetic modulation of treatment responses.
• Facilitates deeper insights into the molecular mechanisms underlying G×T interactions.