guacamol for tdc

limited oracle evaluation

modify the function generate_optimized_molecules(.., scoring_function, ..)

"max_oracle_num = 10000" in scoring_function.py

best_from_chembl --- done 

graph_ga ---- done 

graph_mcts --- done, but confused about max_children 

random_smiles_sampler --- done (no change)

smiles_ga --- done 

tested oracle

1. default='v3'

2. goal_directed_suite_V in ./guacamol/

docking oracle

count docking number

*.py

run

guacamol

oracle init and call: guacamol/guacamol/common_scoring_functions.py

guacamol baseline

(1) SMILES-LSTM guacamol_baselines/smiles_lstm_hc/goal_directed_generation.py (2) Graph-GA guacamol_baselines/graph_ga/goal_directed_generation.py

local

cd /Users/futianfan/Downloads/spring2021/guacamol_tdc
./upload.sh 

server

install guacamol

cd /project/molecular_data/graphnn/guacamol_tdc/guacamol 
python setup.py install 

graph ga

source activate guacamol_baselines
export PATH=$PATH:/project/molecular_data/graphnn/docking/ADFRsuite_installed_directory/bin
export PATH=$PATH:/project/molecular_data/graphnn/docking/autodock_vina_1_1_2_linux_x86/bin
cd /project/molecular_data/graphnn/pyscreener 

python graph_ga_run.py  

smiles lstm

cd /project/molecular_data/graphnn/guacamol_tdc/guacamol 
python setup.py install 


source activate guacamol_baselines
export PATH=$PATH:/project/molecular_data/graphnn/docking/ADFRsuite_installed_directory/bin
export PATH=$PATH:/project/molecular_data/graphnn/docking/autodock_vina_1_1_2_linux_x86/bin
cd /project/molecular_data/graphnn/pyscreener 

python run_smiles_lstm_hc.py  

result analysis

cd /Users/futianfan/Downloads/spring2021/guacamol_tdc
./download.sh 
python results/analysis.py 

pls see results/smiles_lstm_hc.png for results.