François Bérenger
François Bérenger
you can try the correction I suggested. If it works, propose a patch.
Could the package be parameterized by a distance function? I.e. even if the points are high-dimensional, I provide a distance function which behaves well on them and the calculation should...
I completely second @davidegraff: maybe use 50k or 100k as the default limit. If the dataset is bigger than that, automatically use --no_cache_mol. You might also want to make this...
Can this be made available through the CLI? It seems required for very large datasets.
Ok, do you want me to share a 10M public dataset w/ you so that you can reproduce the problem?
~10M molecules; classification setting
Maybe this one also: paper: https://www.nature.com/articles/srep46710 software: https://github.com/oddt/rfscorevs
RF-score by Pedro Balester is also interesting http://chemistry.st-andrews.ac.uk/staff/jbom/group/RF-Score.html I think it is open-source also
Which version of tmap are we supposed to use?
Can't this command: ``` pip install git+https://github.com/reymond-group/[email protected] ``` be made to pull all the required dependencies at an adequate version?