Updates to MST

[nickjcroucher]

Minor updates:

• MST plots now work with lineages
• --previous-clustering now requires a specific file in both visualise and plot_mst modes
• test updated accordingly

Not sure about:

• as sparse_mst.py still requires --distances, do we still require calculation of all-v-all, or are we just using the "metadata" without calculating the values?

[nickjcroucher]

(I don't know why endswith() works, where the regex was failing - could not figure this out).

[johnlees]

Thanks for these fixes/improvements. A couple of suggestions on the code.

Also:

• The tests appear to be failing
• The distances in sparse_mst just load the sample names from the .pkl file and not the dense matrix. But really these should be optional, and only required if the clusters are provided (as it is valid to not colour the nodes by strain). Maybe you could make that change? Or alternatively, can we expect the sample names to be stored as a vertex property?

[johnlees]

Can we also pin the version of graph-tool, had a few reports of that going wrong.
i.e. modify environment.yml to:

graph-tool >=2.35

[nickjcroucher]

I think these are already incorporated, but approving again just in case.

[johnlees]

Can you also add to the environment.yml cugraph and cudf (I think you would also need to add nvidia to the channel list at the top of the file)

[nickjcroucher]

Added cugraph and cudf to environment.yml in 3e5160e.