We are trying to determine why the genes are clustered the way they are when we use the HCL option. What criteria are used and is there anywhere to see this data in the MeV program?
Could you be more specific? Is there some behavior you are expecting that you are not seeing? The distance metric and linkage are user-specified in the initialization dialog. You can save the tree structure and view the data through the right-click option on any HCL tree.
Hi, me too I'm interested in this. I like MeV a lot and really very useful, and I'm thinking to go in it in more deep, and possibly contribute(unfortunately I'm a C, not Java, guy).
I initially thought that MeV always used R libraries, but in the source I find that algorithms are all rewritten and are original in MeV. This is ok, algorithms were not invented by R guys, but of course in R are "reference for all" (I would like a lot a well working R interface that I was not able to setup in my MeV..) But, generally speacking (HCL, SAM, etc), did you "cross-translate" R implementations or create new code directly from original algorithms?
In specific, in MeV HCL, I found strange behaviors in ordering (it seems a strange ordering) and found also differences compared with HCL in MeV scripts, that also do strange ordering and also different clustering using the same data. It seems to me that like some data param in scripts is actually different (defaults?), for sure final normalization is not performed and cluster are different. Worse, in script HCL behavior changes something after the first run and loses some param… using 4.6.2 on win32..