PAIRWISE

PAIRWISE is an all-in-one package for drug synergy prediction. This package allows the user to conduct standardized experiments to compare the prediction performance between reviewed methods.

The user can freely include new datasets, and select preferential cell/drug features to train the deep learning model.

---

Installation

# Unzip 
unzip pairwise.zip
cd pairwise/

#create conda environment
conda env create --name pairwise --file=environment.yml
conda activate pairwise
#To install for PyTorch 1.10.0, simply run on your mac
pip install torch-scatter -f https://pytorch-geometric.com/whl/torch-1.10.0+${cpu}.html
pip install torch-geometric -f https://pytorch-geometric.com/whl/torch-1.10.0+${cpu}.html
pip install torch-cluster -f https://pytorch-geometric.com/whl/torch-1.10.0+${cpu}.html 
pip install torch-sparse -f https://pytorch-geometric.com/whl/torch-1.10.0+${cpu}.html 
pip install torch-spline-conv -f https://pytorch-geometric.com/whl/torch-1.10.0+${cpu}.html

#install pairwise
pip install -e .

#Please download data/ and put it in the same path as setup.py
[data folder] https://drive.google.com/drive/folders/1Uu0YZSxX8GQtV_4ZJmsMrbanmse-Dq6n?usp=sharing/ 

If you are using Mac M1 chip, we recommend checking out this github issue for installation of required dependencies

---

Getting strarted

 python pairwise/main.py --model 'deepsynergy_preuer' --synergy_df 'p13' --train_test_mode train

Features explained

Model	Input feature format		Feature encoders		Features concatenated		Drug1 and drug2 summed
	Cell line	Drug	Cell line	Drug	Cell line	Drug
PAIRWISE	exp	Chemical structures, Drug-target interaction from DrugTargetCommons v2.0	Autoencoders	Pretrained foundation model, DNN			False
ML approaches: LR,RF,XGBoost,ERT	exp or cnv or mut	Drug-target interaction					True
DeepSynergy	exp	Drug chemical descriptor or fingerprints	DNN	DNN			True
MatchMaker	exp	Drug chemical descriptor or fingerprints	DNN	DNN			False
Multitask_DNN	exp	Morgan or MACCS fingerprints, Drug-target interaction	DNN	DNN		False	False
DeepDDS	exp	SMILES2Graph	MLP	GCN			False
TGSynergy	exp	SMILES2Graph	GCN	GCN			False
TranSynergy	exp	Network propagated Drug-target interaction or morgan_fingerprint,smiles,smiles2graph	Transformer	GCN(RWR)+Transformer			False
GraphSynergy	cell_protein,PPI network	drug_protein,PPI network	GCN	GCN			False

---

Data downloaded

PAIWISE used multi-omics datasets.

1. We have provided a cleaned benchmark synergy truset. For details of reporducing, please go to trueset_generation/ to follow the instructions.
2. CCLE dataset including exp, cnv, mut
3. Drug-target interaction dataset from DrugComb, and structures.sdf which enables fingerprints calculation or smiles2graph Link and please put into Data/ folder

---

Models included

In detail, the following drug synergy prediction models were implemented.

• Baseline machine Learning models (random forest, extreme gradient boosting, extremely randomized tree, logistic regression)

• End-to-end deep learning models

  • [1] Kristina Preuer, Richard PI Lewis, Sepp Hochre-iter, Andreas Bender, Krishna C Bulusu, and G ̈unter Klambauer.DeepSynergy: Predicting Anti-Cancer Drug Synergy with DeepLearning.Bioinformatics, 34(9):1538–1546, 2018.
  • [2] Kuru Halil Brahim, Oznur Tastan, and Ercument Cicek. MatchMaker: A Deep Learning Framework for Drug Synergy Prediction. IEEE/ACM Transactions on Computational Biology and Bioinformatics, 2021.
  • [3] Yejin Kim, Shuyu Zheng, Jing Tang, Wenjin Jim Zheng, Zhao Li, and Xiaoqian Jiang. Anticancer Drug Synergy Prediction in Understudied Tissues Using Transfer Learning. Journal of the American Medical Informatics Association, 28(1):42–51, 2021.
  • [4] Jinxian Wang, Xuejun Liu, Siyuan Shen, Lei Deng, and Hui Liu. DeepDDS: Deep Graph Neural Network with Attention Mechanism to Predict Synergistic Drug Combinations. Briefings in Bioinformatics, 09 2021
  • [5] Yiheng Zhu, Zhenqiu Ouyang, Wenbo Chen, Ruiwei Feng, Danny Z Chen, Ji Cao, Jian Wu, TGSA: protein–protein association-based twin graph neural networks for drug response prediction with similarity augmentation, Bioinformatics, Volume 38, Issue 2, 15 January 2022, Pages 461–468
  • [6] Liu, Qiao, and Lei Xie. "TranSynergy: Mechanism-driven interpretable deep neural network for the synergistic prediction and pathway deconvolution of drug combinations." PLoS computational biology 17.2 (2021): e1008653.
  • [7] Yang, Jiannan, Zhongzhi Xu, William Ka Kei Wu, Qian Chu, and Qingpeng Zhang. "GraphSynergy: a network-inspired deep learning model for anticancer drug combination prediction." Journal of the American Medical Informatics Association 28, no. 11 (2021): 2336-2345.

---

Customized dataset

Use customized dataset to test. The testing drug combos are sourced from specialized tissues. The testing results are stored in /results/predicts_"Model"_"Customized".csv

---

License

MIT