François Bérenger
François Bérenger
if MinHash is so important to map4, it should probably be included and not be provided by an external dependency
maybe automate via a script the calculation of phys-chem properties for the ChEMBL 100k subset and generated molecules from it; then calculation of histograms for each property
generate a large molecular sample using uniform random, see what happens in the profile
same for TS mode
done for TS mode; nothing special to report about (we call a lot Gauss.sample if there are many fragments to choose from; which is expected)
it creates a Ht of smi2frag_id
this is in place and working
(shell/AWK/Perl-like scripting in OCaml)
liblinear is a very famous kind of chainsaw for machine-learning work
related to https://github.com/UnixJunkie/svmwrap/issues/12