unixjunkie / acpc Goto Github PK

gnuplot
pip install CROC
pip install sympy

at least during parsing of MOL2 files
there is some work saved already in the branch https://github.com/UnixJunkie/ACPC/tree/no_hydrogens

this is for open babel

the fix would be to use the unix sort command on the file instead

cf. references in JCAMD 2016 30:209-217

a functor is needed, parameterized by a module providing
a tanimoto fun and a molecules list

compute then dump out the full distance matrix in csv format
so that we can import it into R.

with Parmap of course

Cf. le commentaire de CM pendant ma soutenance.

There are two other properties that we could use instead of atomic radius
in order to improve utility in the steric space.
Cf. my research_ideas file.

We need to add a new command --dump out_file.
It should write one line per compound: all autocorr values from -max_dist (for the negative values)
to +max_dist (for the positive values).

Cf. "Autocorrelation descriptor improvements for QSAR: 2DA_Sign and 3DA_Sign".
JCAMD 2016 30:209-217

after #8 is done

based on the presence of their name in a hash table.
This hash table should be populated at startup if the user provided a filename
(as an option) where to read the name of active molecules from.
This is way better than having to force people to name their molecules in a specific way.

branch 1.2 can compile with 4.06.1 and recent opam packages

use Paretto ranking, BM clustering and play with a.
A change of the kernel function might also bee needed but this would have a huge impact
on performance.