Version 3.4 
- adds a new function to compute the posterior distribution of the 
  sampled SNPs. It is called imputedGenotypes

Version 3.5
- adds interactive and batch parallel functions for variableSelector

REMINDERS:

(1) Are you using the right .Rprofile file? (BOTH)
(2) Did you remember to load the BAMD library with lib.loc argument (BATCH)
(3) Did you remember to put mpi.quit()? (BOTH)
(4) Did you remember to close the slaves with dellog=FALSE? (INTERACTIVE)

To run the batch parallel variable selector:
    copy RprofileBATCH to your working directory as .Rprofile, and do 
    something like:

    $ mpirun -np 3 -machinefile $PBS_NODEFILE R --no-save -q CMD BATCH \
    > BAMD_3-5/demo/demo1.R /dev/null

To run the interactive parallel variable selector:
    copy RprofileINTERACTIVE to your working directory as .Rprofile, and do:

    $ mpirun -np 3 -machinefile $PBS_NODEFILE R --no-save -q

    The above will start the slaves, and give you an interactive session on 
    the master. Then you can run the appropriate commands from demo1.R 
    line by line. (load library BAMD, call gibbsSampler, call 
    variableSelectorInteractP, then close slaves and quit).
