fasmifra

Molecular Generation by Fast Assembly of SMILES Fragments
README

Reference implementation for the article
"Fast Molecular Generation by Assembly of (Deep)SMILES Fragments".
Generate molecules fast from a molecular training set while also
doing training-set distribution matching.

Installing the software

On Linux, you need python3-rdkit to be installed.

For OCaml programmers, you can clone this repository
then type 'make && make install'.
Note that you need to have opam installed and configured.

For end-users:

sudo apt install opam
opam init
eval `opam config env`
opam install fasmifra

For programmer end-users, or if the opam package is not ready yet:

sudo apt install opam
opam init
eval `opam config env`
opam pin add fasmifra https://github.com/UnixJunkie/FASMIFRA.git

We are currently working on an automated self-installer; stay tuned.

Fragmenting molecules

Those molecules are your "molecular training set".

fasmifra_fragment.py -i my_molecules.smi -o my_molecules_frags.smi

If you fragment rather small molecules, you might want to use the -w option
and pass a smaller recommended fragment weight than the default (150 Da).

usage: fasmifra_fragment.py [-h] [-i input.smi] [-o output.smi] [--seed SEED]
                            [-n NB_PASSES] [-w FRAG_WEIGHT]

fragment molecules (tag cleaved bonds)

optional arguments:
  -h, --help      show this help message and exit
  -i input.smi    molecules input file
  -o output.smi   fragments output file
  --seed SEED     RNG seed
  -n NB_PASSES    number of fragmentation passes
  -w FRAG_WEIGHT  fragment weight (default=150Da)

Generating molecules from fragments

fasmifra -n 100000 -i my_molecules_frags.smi -o my_molecules_gen.smi
usage:
  fasmifra
  -n <int>: how many molecules to generate
  -i <filename>: smiles fragments input file
  -o <filenams>: output file
  [--seed <int>]: RNG seed
  [--deep-smiles]: input/output molecules in DeepSMILES no-rings format

Bibliography

[1] Berenger, F., & Tsuda, K. (2021).
"Fast Molecular Generation by Assembly of (Deep)SMILES Fragments".
Submitted manuscript.

[2] O'Boyle, N., & Dalke, A. (2018).
"DeepSMILES: an adaptation of SMILES for use in machine-learning of chemical structures".
chemrxiv.org

[3] Weininger, D. (1988). SMILES, a chemical language and information system.
"1. Introduction to methodology and encoding rules".
Journal of chemical information and computer sciences, 28(1), 31-36.

Install
Sources
v1.1.0.tar.gz
md5=d12b87ec8a1a361ef4d9fa1c23615869
Dependencies
conf-python-3
>= "1.0.0"
line_oriented
>= "1.0.0"
parany
>= "12.0.0"
minicli
>= "5.0.0"
dune
>= "1.11"
dolog
>= "6.0.0"
batteries
>= "3.3.0"
Reverse Dependencies