Switch/Update to other HUNTRESS Implementation

[gordonkoehn]

It has been discovered that the scphylo implementation does fail silently when running it with multiple threads as essential for larger cell-mutation profiles.

This has been confirmed by the developers.

• Recommended Resources to use HUNTRESS / multiple thread usage leads to halt of HUNTRESS ?  faridrashidi/scphylo-tools#10 (comment)

The issue has been addressed by the developers - and we shall update our scphylo dependency.

• check why @pawel-czyz did a fork to start with / Why did we remove the numba dependency?

[gordonkoehn]

I'll try to just update the package for now.

[gordonkoehn]

Just updating to the newest version of scyphlo works when downgrading nextworkX, but I get trouble with numpy and numba - I am not sure why, something about a long not being defined.

Perhaps passing variables back to my numpy environment brings trouble - really don't know. Let's try to do a fork like @pawel-czyz did on my own, removing netwrokX and numba again.

[gordonkoehn]

Make my own fork removing mamba and unpinning the networkX version. All tests work.

• https://github.com/gordonkoehn/scphylo-tools

[gordonkoehn]

HUNTRESS indeed seems to work for multiple threads here.

• Test 8 cores 16 GB ram locally for 1000 cells and 10 mutations.

[gordonkoehn]

Merge before testing. - Try in HPC EULER.