I've been testing out the pamr package and have read the paper "Diagnosis of multiple cancer types by shrunken centroids of gene expression" by Tibshirani, et. al. In the paper it mentions using 10-fold cross-validation. The pamr.cv function defaults to setting nfold to the smallest class size, in this case 8. Using the khan.txt tutorial data I've tried setting nfold = 10 to ensure I can reproduce the results from the paper. It still only does an 8-fold cross-validation. Is this a bug or am I missing something? I'm using pamr ver 1.31 and R ver 2.6.1. - Pat ________________________________________________________________ ____________________ Looking for last minute shopping deals?