To obtain reliable results, we randomly divided the study population into two sub-samples of equal size which were used for screening and validation purposes respectively (see Materials and Methods). In these analyses, each risk factor was considered separately whereas all expression traits were envisaged jointly for their potential association with the risk factor, considered here as the dependent variable. The screening/validation procedure was repeated 250 times and for each risk factor, we report expression traits associated (P<0.01) with the risk factor in more than 25% of the replicates. This stringent approach led to the identification of 106 independent expression correlates for the ten risk factors ( Table 6 ), a much reduced number compared to the 1,662 expression traits previously identified by the one-to-one association analysis presented in Table 5 .
