Hi, recently maybe I found two possible bugs in exonmap. When use splanova, according to exonmap tutorial and papers, only exonic probesets should be picked first. Then translate them to genes and submit genes to splanova. Here is the problem. The genes may have only one exonic probeset and other non-exonic probesets. In .gene.midas, non-exonic probesets would be removed, and only one exonic probeset would induce error in function(x). The second is still about .gene.midas. Following line in the function defines to start with the second col. colnames(out) <- colnames(pf)[2:(length(colnames(pf)) - 1)] Why from the second? I guess maybe the first col is assumed to be the base group in midas. Then there will be wrong col name in output of fval if other col is set to be base group. By the way, it seems fval output is p val. Hope these may help. Xin