I have noticed an error when using MeV 4.7.4 on a Win7 machine. When trying to perform an HCL on a geneset (2310 genes; 3 samples) it takes 5 minutes to do the calculation and the HCL tree is in the original unclustered configuration. The data works fine on another machine with WinXP, TMEV 4.3 and Java 1.6.0_30.
Since this is a new set-up this year (Win7 upgrade and TMEV upgrade) where should I be looking for the error?
I have been playing with OS version/MEV version combinations on several lab setups to try and resolve the HCL problem I noted above.
Even on XP once the MEV version goes to 4.6, the HCL stops working properly with even a smaller subset of the data I have been working with. Versions 4.3 to 4.5 all work the same under XP, when I move to version 4.6 and beyond the HCL tree is displayed unsorted and the tree branch lengths are all the same size.
What happened to the HCL algorithm between MEV 4.5 and 4.6?
I would like to help, but unfortunately I'm not seeing the same behavior. Using my datasets, on XP, MeV seems to cluster identically across each version, unless the order optimization is used. There does seem to be something strange in the branch length in 4.6, but is fixed by 4.7.
Are you using any filtering? Is there an error message in the console? Could you give us an example dataset that could be used to reproduce this issue?