CIGER - Chemical-induced Gene Expression Ranking

CIGER - Chemical-induced gene expression ranking and its application to pancreatic cancer drug repurposing

---

Code by Thai-Hoang Pham at Ohio State University.

1. Introduction

This repository contains source code (CIGER) and data for paper "Chemical-induced gene expression ranking and its application to pancreatic cancer drug repurposing" (Patterns 3 (2022))

CIGER is a Python implementation of the neural network-based model that predicts the rankings of genes in the whole chemical-induced gene expression profiles given molecular structures.

The experimental results show that CIGER significantly outperforms existing methods in both ranking and classification metrics for gene expression prediction task. Furthermore, a new drug screening pipeline based on CIGER is proposed to select potential treatments for pancreatic cancer from the DrugBank database, thereby showing the effectiveness of CIGER for phenotypic compound screening of precision drug discovery in practice.

2. CIGER architecture

Figure 1: Overall architecture of CIGER

3. Pipeline

Figure 2: Drug screening pipeline using CIGER. This model is trained with the LINCS L1000 dataset to learn the relation between gene expression profiles and molecular structures (i.e., SMILES). Then molecular structures retrieved from the DrugBank database are put into CIGER to generate the corresponding gene expression profiles. Finally, these profiles are compared with disease profiles calculated from treated and untreated samples to find the most potential treatments for that disease.

4. Installation

CIGER depends on Numpy, SciPy, PyTorch (CUDA toolkit if use GPU), scikit-learn, tqdm, and RDKit. You must have them installed before using CIGER.

The simple way to install them is using conda:

	$ conda install numpy scipy scikit-learn rdkit pytorch tqdm

5. Usage

5.1. Data

The datasets used to train CIGER are located at folder CIGER/data/

5.2. Training CIGER

The training script for CIGER is located at folder CIGER/. The Python script is train.py. Example of running this script is in train.sh.

    $ cd CIGER
    $ bash train.sh

Important Arguments:

--fp_type: Chemical representation method. Select between ECFP and neural FP.

--label_type: Gene expression label used for training. Select from real, real reverse, binary, and binary reverse. real: Training with gene expression values (z-scores). real reverse: Training with reversed gene expression values. binary: gene expression values are converted to binary values with top 95th percentile values as positive label. binary_reverse: gene expression values are converted to binary values with bottom 5th percentile values as positive label.

--loss_type: Learning-to-rank objective function. Select from pairwise_ranknet, list_wise_listnet, list_wise_listmle, list_wise_rankcosine, list_wise_ndcg.

Scripts for pancreatic cancer are located at folder CIGER/drug_repurposing/

    $ cd CIGER/drug_repurposing
    $ python drug_repurposing_precision.py # using precision score
    $ python drug_repurposing_gsea.py # using enrichment score

6. References

Pham, TH., Qiu, Y., Liu, J. et al. Chemical-induced gene expression ranking and its application to pancreatic cancer drug repurposing. Patterns 3, (2022).

7. Contact

Thai-Hoang Pham < pham.375@osu.edu >

Department of Computer Science and Engineering, Ohio State University, USA