Working_with BEAST_on_ATLAS.txt

preamble on BEAST v. BEAST2: For almost everything you would want to do, BEAST2 should be preferred. 
BEAST can run some older models that their creators never ported over to BEAST2, 
but BEAST2 is meant to be a full replacement of BEAST, according to its authors
(see: https://www.beast2.org/BEAST/ and https://www.beast2.org/features/)

Installing BEAST and Beagle on Atlas:
	The best approach (in my opinion) for installing BEAST and Beagle on an HPC can be found here:

	http://felixgrewe.github.io/linux_cookbook/install_BEAST2.html
	
	To be comprehensive about your options, you have three ways to get up and running:
	
	
	1) Use miniconda (works, but only has version 2.6.3 or earlier of BEAST):
	
		# load conda	
		module load miniconda3
	
		# create new conda environment and install beast
		conda create -n beast beast2 beagle-lib 
		
		# load java
		module load openjdk

	
	2) Use version of BEAST on Atlas (works, but only has version 2.7.4 of BEAST, and you still need conda for beagle)

		# This should also load Java, so no need to do that separately 
		Module load beast2

		# create new conda environment and install beast
		conda create -n beast beast2 beagle-lib 

	
	3) Download any version of BEAST you want from the GitHub page (Again, this is the  best approach I think)
	
		# in the directory you want BEAST to live, run the following, replacing the version number of your choice:

		wget https://github.com/CompEvol/beast2/releases/download/v2.7.7/BEAST.v2.7.7.Linux.x86.tgz

		# untar 
		tar zxvf BEAST.v2.7.7.Linux.x86.tgz

		# again, we need java
		module load openjdk
	
		# and Beagle:
		conda create -n beagle beagle-lib
	

All of these methods will install BEAST and BEAGLE, but any packages the you need for fancy models you 
will be running need to be installed still using the built in package manager for BEAST. 
See here for reference, under the section “Installation on cluster”:

	https://www.beast2.org/managing-packages/