There have been a couple quick updates to the FATCAT+3dMVM tools.

Firstly, one can now specify a subset of the parameters to include in the set of 3dMVM model runs in fat_mvm_scripter.py. Each *.grid or *.netcc file (particularly the former) can contain several parameter matrices, and it may be inconvenient to run them all at a single go. For example, one might want models of FA, MD or NV parameters but not L1, RD, et al. from the *.grid files. Therefore, one can now input (via commandline or text file) a list of parameters to model with the variable set (see 'fat_mvm_scripter -h' for the parameter-selecting options).

Additionally, there has been a small bug fix in fat_mvm_prep.py in dealing with missing data in the group's CSV file. It is now possible to include missing data (marked with 'NA') in the input CSV file-- and fat_mvm_prep.py will warn you about this by default. Note, however, that 3dMVM still does not enjoy getting NAs. *Currently*, it will remove those subjects from the modeling (i.e., no partial data allowed). *Potentially* in the near future, something like LME-like partial data analysis will be allowed. Watch this space.

Finally, a draft of a recent study using the FATCAT+3dMVM tools has been accepted in HBM, and it can be found for reading here:
afni.nimh.nih.gov/pub/dist/papers/TaylorEtal14_DTI_PAE.pdf
It gives a practical example of a group study that statistically models a network of diffusion parameter values with subject data (in this case, investigating associations of prenatal alcohol exposure with white matter properties of newborns, while controling for several factors such as infant age, etc.).

Cheers again,
pt, Gang, et al.