To convert the files to MrBayes format I generally use Mesquite (http://mesquiteproject.org/mesquite/mesquite.html). Simply import your alignment, the program takes most formats, and then export as nexus for MrBayes. However, in relation to the Bayes block this is dependent on your input (DNA/amino acid), the model of evolution (GTR, HKY etc), how many runs you want, the number of hot/cold chains etc etc etc. To learn to make a bayes block your best port of call is without doubt the MrBayes manual. Here you'll find lots of examples as well! 

good luck. 

To convert the files to MrBayes format I generally use Mesquite (http://mesquiteproject.org/mesquite/mesquite.html). Simply import your alignment, the program takes most formats, and then export as nexus for MrBayes. However, in relation to the Bayes block this is dependent on your input (DNA/amino acid), the model of evolution (GTR, HKY etc), how many runs you want, the number of hot/cold chains etc etc etc. To learn to make a bayes block your best port of call is without doubt the MrBayes manual. Here you'll find lots of examples as well! 

good luck. 

I wrote a program called Particular a couple of years ago that can create a MrBayes command files from an alignment. It's main feature is that it can properly set up partitions and can handle different evolutionary models for the partitions. No guarantees as to how bug-free it is, but it worked for me and it can save you a lot of time.

https://sites.google.com/site/dojocafe/

Though late....

nexus files can be easily created via online tools i.e. Readseq and Seqret available at the EMBL site. Define your output as PAUP nexus/Interleaved/non-interleaved