Hi, I found two sorts of functions to use the SOM algorithm, one in the package "som" and the other one in the package "class", but none of them deal with the missing values. The som function in package "som" gives results with option init="random", but the value of the code vector has missing values where at least one of the training vectors has a missing value which is not acceptable to me. I know SOM algorithm is quite good to handle missing values, does anyone know if a som function dealing with missing values exists and where I can find it. I heard about "GeneSOM" package but couldn't find it on the CRAN website, is it the same as "som" package? Thanks for any help! Eric ESPOSITO