During revision of this manuscript, another paper appeared,50 which similar to our paper compared the results of an eQTL association analysis in human livers (two collectives with n1=206 and n2=60) with the Seattle study.16 Using different validation sets, they found replication rates between 49 and 58% for cis-eQTLs, that is, about half of all cis-associations failed to be validated in their study. A thorough investigation of the factors influencing reproducibility revealed similar reasons as those mentioned in our study. The lower replication rate in our study compared with50 may be explained by more relaxed validation criteria in their study, in particular regarding muss less stringent multiple correction conditions. Other likely reasons for differing replication rates include the factors discussed above, that is, origin and sampling conditions of the tissues, statistical modeling, as well as ethnicity and other covariates of the liver donors.
