The relationship between gene expression and each CHD risk factor was tested by a linear regression model using R-lm, with gene expression as the dependent variable. Association with age was adjusted for gender while association with other risk factors were adjusted for gender, age and, if specified, BMI. A square root transformation was applied to CRP levels to remove positive skewness. A study-wise statistical significance threshold of 3.9×10−7 was used to correct for the number of tests (10 risk factors ×12,808 gene expressions). For each expression trait that was associated with a risk factor and also affected by cis eSNPs, we tested the interaction between the risk factor and the best cis eSNP on expression in a regression model.
