Welcome to AntimicroBial compounD Prediction Server abbreviated as "ABDPred". The server utilizes four machine learning (ML) classifiers - XGBoost (XGB), Random Forest (RF), Gradient Boosting (GBC), and Deep Neural Network (DNN) - to predict antimicrobial compounds, and then aggregates all prediction results through the soft voting technique.
The server can predict whether the input chemical/drug/compounds might act as an antimicrobial compound with a sensitivity value of 68.75%, specificity value of 92.50%, accuracy value of 80.62% and Matthews correlation coefficient: MCC value of 60.12%(as tested on the blind dataset). The ‘drug-likeness and bioavailability profile’ of the input chemical/compounds would also be checked in the server. Additionally, the server can compare the two-dimensional (2D) structures of our datasets based on the Tanimoto Coefficient Similarity score. The web framework of the server has been designed using PHP 7.3.28 (cli), Python (3.8.6), and Apache Web Server on CentOS 7.

Users can type the PubChem CID of the query chemical/drug/compound in the green colored text field. An example of a known antibiotic drug named "Gramicidin D" (45267103), has been given.

 For more information about ABDPred server background and usage of the server please visit the “About” page and “Help” page.

 The datasets, source code of ABDPred sever, and source codes of model development can be found at the Download page

"ABDpred: Prediction of active antimicrobial compounds using supervised machine learning techniques", By Jana Tanmoy et al . is accepted on Monday, 26 June, 2023 for publication in the Indian Journal of Medical Research journal (Impact Factor: 5.274).

ABDpred is also available at https://github.com/janatanmoy5/abdpred