- Download the data from NCBI in fasta format (its an option in one of the pull down headers).
- Load it into the latest version of MEGA, version 7(?), perform the alignment therein.
- Export from MEGA as Nexus format and upload it into Beast
You could of course perform the alignment in Clustalx, Clustalo or Muscle or T-Coffee, (hmmm I could write an alignment program just to name it T-4two (sorry British joke)). If the output option of "nexus" was not present in the alignment program of your choice, then again upload the alignment into MEGA and export it as nexus.
Anyway, I could go into specific detail about the various points and clicks therein, but I think its easy to figure out. Most phylogeny programs will do an automatic fasta->nexus conversion, but anyway.
Its easy, Beast is not trivial to use properly however and ensuring convergence is not at all easy, in addition the multiple simulation a priori models, whilst powerful, give different answers. To try and explain this a different way, it is easy to get an answer in Beast, but difficult to get a robust answer from it and thats the bit I would give serious consideration about. It does amaze how widespread the program is (I guess its the nice GUI), given how complicated it is to obtain robust result.
Just to add you could try the online tool here for the fasta to nexus conversion: . I've not tried it, its just an after thought. If you do Python, Biopython will do it, but its just as easy to code.