harisankarsadasivan / lora Goto Github PK

To get a sinlge instance of a phylogenetic tree: 'python2 run.py <args>'
    - This will print the correlation coefficient to the terminal and save a .png
    file of the dendrogram.
To generate all combinations of embeddings for a given dimensionality d: './run_all.sh d'
    - This will print all correlation coefficients to the terminal

Note: run.py will first check if the embeddings are saved in the data directory.  If they are not present,
they will be generated which can take several hours, depending on the parameters.  If the embeddings are present,
all pairwise distances are calculated for the loaded embeddings and the dendrogram is constructed, which takes on
the order of seconds.

If you intend to load embeddings from memory, convenience scripts are provided to create and load a virtual
environment on flux.  Running create_virtual_env.sh, then load_virtual_env.sh, then install_dependencies.sh
should satisfy all prerequisites to run the main script (we are fairly confident that we have caught all
dependencies, but additional installs may be required).


For input data generation, use bbmap (https://github.com/BioInfoTools/BBMap) and Flye (https://github.com/fenderglass/Flye). Flye needs to be tweaked 
to skip the polishing and final steps to just produce repeat graphs.