brsynth / RetroPathRL

Reinforcement Learning based bioretrosynthesis tool
MIT License
48 stars 17 forks source link

Unable to set up repo because of missing rule files #15

Closed AlejandroSanchezCano closed 1 year ago

AlejandroSanchezCano commented 1 year ago

Good afternoon.

I have gone through the installation guide. Namely, I have run:

conda create --name MCTS python=3.6
conda activate MCTS
conda install --channel rdkit rdkit=2019.03.1.0
conda install pytest
conda install pyyaml

pip install -e .

But when I want to run:

python calculate_rule_sets_similarity.py --rule_address_with_H your_rule_address --rule_address_without_H your_rule_address

It gives me an error because your_rule_address, as the help manual for the function indicates is the rule file with(out) hydrogen that will be used. What is this file? Are these the chemical rules that RetroPathRL uses? If so, am I supposed to get them myself? It is not very clear how to proceed. Please I would appreciate if somebody found lend me a hand here.

tduigou commented 1 year ago

Hi @AlejandroSanchezCano,

Indeed, these rule_address_with[out]_H files are the files providing the reaction rules (i.e. the chemical rules) that RPRL will use. They can be downloaded from the RetroRules database at https://retrorules.org/dl (at the bottom of the page, there are two datasets for RetroPath RL, one with explicit Hs, another without).

I updated the README file to clearly state where to find the files.