confusedant/ctoxpred2

Comprehensive cardiotoxicity prediction tool of small molecules on three targets: hERG, Nav1.5, Cav1.2

CToxPred2

Comprehensive cardiotoxicity prediction tool of small molecules on three targets: hERG, Nav1.5, Cav1.2

❗Clone first the whole repository package and follow the steps bellow.

Prerequisites

1- Create and activate a conda environment:

	$conda create -n ctoxpred2 python=3.9
	$conda activate ctoxpred2

2- Install packages:

	$bash install.sh

3- Clone the repository:

	$git clone [email protected]:issararab/CToxPred2.git

4- Move to the repository:

	$cd CToxPred2

5- Start the App:

	$python app.py

❗Before running the tool (step 5), ensure to decompress the models located under "CToxPred2 > models".

• To execute screening/predictions from a notebook, please consult the example provided at: notebooks/analysis_notebook.ipynb
• To recreate figures as published in the paper, start by decompressing the result CSV files under "CToxPred2/data/analyzed_sets". Then, utilize the notebook located at: notebooks/make_predictions.ipynb
• The software facilitates users to export the results as a CSV file and store them in a folder of their choosing.

Manuscript

TBA

Data availability

To re-train the models, re-evaluate the models using the same test sets, or re-run the analysis notebook, you will find all the data in the folder './data'. To access the complete library store containing approximately 2 million small molecules, including the entire ChEMBL database, and their corresponding feature representations, retrieve the publicly available database from Zenodo.

Hot stuff

• Evaluation of the CToxPred performance compared to CardioGenAI, trained and tested using the same data, where (a) is performance on Test-70 and (b) performance on Test-60.

• t-SNE visualizations showing the distributions of the labeled and unlabeled molecules in the development set and the two external test sets (Eval-60 and Eval-70) for (a) hERG (b) Nav1.5 and (c) Cav1.2.