#MultiTox

Title: MultiTox: Computational Ensemble Stack Model To Predict the Multiclass Toxicity of Proteins

Authors: Harshika Sharma, Mayank Singh Thakur, Avinash Barala, Modh Saad Khan, Ganesh Bagler.

Highlights:

• MultiTox predicts the multiclass toxicity of proteins based on their mechanism of action in humans.

• The model is trained on a large dataset of 20,029 toxins and 4,245 non-toxins.

• It classifies proteins into neurotoxins, cytotoxins, hemotoxins, and enterotoxins.

• The model is trained by using ESM-2 pre-trained features.

• The model uses a stack ensemble-based approach for the prediction of multiclass toxicity.

• MultiTox achieves state-of-the-art performance, with an accuracy of 91.07%, an F1-score of 90.73, and an MCC of 91.61.