anaeliaovalle / atc-mt-dti

Augmenting MT-DTI model with ATC drug classification

Augmenting MT-DTI model with ATC drug classification for Drug-Drug Chemical Subgroup Similarity

https://www.who.int/medicines/regulation/medicines-safety/toolkit_atc/en/

CS247 Documentation for running the ATC embedding

1. Download data.tar.gz to get the Kiba tfrecords
2. Run the model finetuning on the ATC Embedding

cd src/finetune
export PYTHONPATH='../../'
python finetune_demo.py  --use-atc --save-model-dir ./yes-atc 

2. Run the predictions

For running predictions on ATC embedding with Kiba data:

cd src/predict
export PYTHONPATH='../../'
python predict_demo.py  --use-atc --load-model-dir ../finetune/yes-atc 

For running predictions on ATC embedding with COVID19 data:

cd src/predict
export PYTHONPATH='../../'
python predict_demo.py  --covid --use-atc --load-model-dir ../finetune/yes-atc 

Description of jupyter notebooks

1. src/data_processing_notebooks/'ATC Adjacency Matrix.ipynb'

This notebook was used to prepare the ATC adjacency matrix for the ATC embedding. A database of drugs with ATC codes were used to create an adjacency matrix based on the level 2 ATC classification. 

2. src/data_processing_notebooks/'Data Assembly for COVID.ipynb'

This notebook was used to prepare the appropriate files for running the ATC-MT-DTI model for COVID-19. The files generated are the protein.txt and ligands_can.txt files which are data files for the protein targets and drug compounds repectively.

3. src/data_processing_notebooks/'Mapping ATC drug names to ChEMBL ID.ipynb'

This notebook was used to map the drug names from the ATC database (1st jupyter notebook) to a common naming schema using ChEMBL IDs. This mapping was used to look up ATC embeddings for drugs with known ChEMBL IDs.

4. src/data_processing_notebooks/'ATC Adjacency Matrix.ipynb'

This notebook was used to generate the r-square plot for the KIBA benchmarking dataset and calculates R^2 statistic and mean-square error. An r-squared plot is generated for both the ATC-MT-DTI model and the baseline model. 

5. data/covid/'COVID DTI Prediction Results.ipynb'

This notebook was used to analyze the resulting drug target interactions for the COVID-19 dataset. This notebook reads in the result file and looks up the ChEMBL IDs for each drug to get its preferred drug name as well as its ATC classification if available.

Forked Repo Documentation

---

MT-DTI

An official Molecule Transformer Drug Target Interaction (MT-DTI) model

• Author: Bonggun Shin
• Paper: Shin, B., Park, S., Kang, K. & Ho, J.C.. (2019). Self-Attention Based Molecule Representation for Predicting Drug-Target Interaction. Proceedings of the 4th Machine Learning for Healthcare Conference, in PMLR 106:230-248

Required Files

• Download data.tar.gz
  • This includes;
    • Orginal KIBA dataset from DeepDTA
    • tfrecord for KIBA dataset
    • Pretrained weights of the molecule transformer
    • Finetuned weights of the MT-DTI model for KIBA fold0
• Unzip it (folder name is data) and place under the project root

cd mtdti_demo
# place the downloaded file (data.tar.gz) at "mtdti_demo"
tar xzfv data.tar.gz

• These files sholud be in the right places

mtdti_demo/data/chembl_to_cids.txt
mtdti_demo/data/CID_CHEMBL.tsv
mtdti_demo/data/kiba/*
mtdti_demo/data/kiba/folds/*
mtdti_demo/data/kiba/mbert_cnn_v1_lr0.0001_k12_k12_k12_fold0/*
mtdti_demo/data/kiba/tfrecord/*.tfrecord
mtdti_demo/data/pretrain/*
mtdti_demo/data/pretrain/mbert_6500k/*

VirtualEnv

• install mkvirtualenv
• create a dti env with the following commands

mkvirtualenv --python=`which python3` dti
pip install tensorflow-gpu==1.12.0

Preprocessing

• If downloaded data.tar.gz, then you can skip these preprocessings

• Transform kiba dataset into one pickle file

python kiba_to_pkl.py 

# Resulted files
mtdti_demo/data/kiba/kiba_b.cpkl

• Prepare Tensorflow Record files

cd src/preprocess
export PYTHONPATH='../../'
python tfrecord_writer.py 

# Resulted files
mtdti_demo/data/kiba/tfrecord/*.tfrecord

FineTuning

• If downloaded data.tar.gz, then you can skip this finetuning

cd src/finetune
export PYTHONPATH='../../'
python finetune_demo.py 

Prediction

cd src/predict
export PYTHONPATH='../../'
python predict_demo.py