Unraveling gene-gene interactions regulated by ligands of the aryl hydrocarbon receptor

The co-expression of genes coupled to additive probabilistic relationships was used to identify gene sets predictive of the complex biological interactions regulated by ligands of the aryl hydrocarbon receptor (Ahr). To maximize the number of possible gene-gene combinations, data sets from murine embryonic kidney, fetal heart, and vascular smooth muscle cells challenged in vitro with ligands of the Ahr were used to create predictor/training data sets. Biologically relevant gene predictor sets were calculated for Ahr, cytochrome P450 1B1, insulin-like growth factor-binding protein-5, lysyl oxidase, and osteopontin. Transcript levels were categorized into ternary expressions and target genes selected from the data set and tested for all possible combinations using three gene sets as predictors of transitional level. The goodness of prediction for each set was quantified using a multivariate nonlinear coefficient of determination. Evidence is presented that predictor gene combinations can be effectively used to resolve gene-gene interactions regulated by Ahr ligands.