SYNERGY PREDICTION OF ANTIMICROBIAL PEPTIDES (AMPS) WITH OTHER AGENTS USING SUPERVISED MACHINE LEARNING

Alex Sanchez Yumbo; Thiago Souza; Isabella Alvim Guedes; Dr. Laurent Emmanuel Dardenne; Marisa Fabiana Nicolás

All Papers

Submission

Paper Title

SYNERGY PREDICTION OF ANTIMICROBIAL PEPTIDES (AMPS) WITH OTHER AGENTS USING SUPERVISED MACHINE LEARNING

Authors

Alex Sanchez Yumbo
Thiago Souza
Isabella Alvim Guedes
Dr. Laurent Emmanuel Dardenne
Marisa Fabiana Nicolás

Modality

Poster

Subject area

Database and Software Development

Publishing Date

21/11/2024

Country of Publishing

Brazil | Brasil

Language of Publishing

Inglês

Paper Page

https://www.even3.com.br/anais/xmeeting-2024/832663-synergy-prediction-of-antimicrobial-peptides-(amps)-with-other-agents-using-supervised-machine-learning

ISBN

978-65-272-0843-3

Keywords

Synergism, Antimicrobial Peptide, Supervised Machine learning, LightGBM, PCA

Summary

Combining two or more antimicrobial agents holds promise for exerting a potentiated effect against pathogens, which could lower the risk of drug-resistant subpopulations emerging equally resistant to all drugs. This phenomenon, known as synergism, has emerged as a promising strategy to address multi-drug resistance pathogens. However, experimental validation of potential synergistic combinations in vivo or in vitro may be time-consuming and high cost due to the vast number of possible combinations. Artificial intelligence (AI) offers a solution by leveraging existing knowledge of synergistic pairs to identify promising combinations efficiently. Previous studies have successfully utilized AI models to predict the synergy of combined conventional antibiotics with remarkable accuracy and performance. This method employs classification to categorize combinations as either synergetic or not, allowing for the use of classification algorithms. These models offer the probability of a combination being synergetic. Previous studies have demonstrated the effectiveness of a Light Gradient Boosted Machine (LightGBM), a tree-based supervised machine learning algorithm, for predicting synergism, showcasing high accuracy values. This study explores the synergistic potential of Antimicrobial Peptides (AMPs) when associated with other antimicrobial agents using the LightGBM classifier. Combination data were extracted from the Database of Antimicrobial Activity and Structure of Peptides (DBAASP). This database includes combination entries that involve AMP reporting the Fractional Inhibitory Concentration Index (FICI) as a synergism metric. Chemical information was retrieved from the DrugBank database. We incorporated several features into our analysis, including physicochemical properties and the peptide sequence of AMP, chemical data and the mechanism of action of antimicrobial agents, their Minimal Inhibition Concentration (MIC), and the FICI value. Our study distinguishes itself through its focus on rigorous feature engineering, which sets it apart from other methodologies. While specific studies tackle these steps empirically across the entire dataset, we have adopted a more systematic approach. Specifically, our feature engineering phase incorporates encoding categorical features using One Hot Encoding, implementing data rescaling via Standard Scaler, and conducting correlated feature deletion using Principal Component Analysis (PCA). Additionally, we adhere to best practices by applying these transformations exclusively to the training dataset, thereby reducing the likelihood of data leakage and enhancing the reliability of our model's performance on unseen data. We employed a k-fold cross-validation approach to assess the overall performance of our models, evaluating accuracy and area under the curve (AUC) on unseen datasets. Our model demonstrated an accuracy of 78.04% and an AUC of 0.80. In comparison, prior research has documented an accuracy of 71.74% and an AUC of 0.76 during external model evaluation. These findings underscore the competitive performance of our approach with state-of-the-art models. Our current aim involves analyzing the potential of regression models for this synergism prediction using the FICI index as model outcome prediction.

Title of the Event: 20º Congresso Brasileiro de Bioinformática: X-Meeting 2024
City of the Event: Salvador
Title of the Proceedings of the event: X-Meeting presentations
Name of the Publisher: Even3
Means of Dissemination: Meio Digital

How to cite

YUMBO, Alex Sanchez et al.. SYNERGY PREDICTION OF ANTIMICROBIAL PEPTIDES (AMPS) WITH OTHER AGENTS USING SUPERVISED MACHINE LEARNING.. In: X-Meeting presentations. Anais...Salvador(BA) Hotel Deville Prime, 2024. Available in: https//www.even3.com.br/anais/xmeeting-2024/832663-SYNERGY-PREDICTION-OF-ANTIMICROBIAL-PEPTIDES-(AMPS)-WITH-OTHER-AGENTS-USING-SUPERVISED-MACHINE-LEARNING. Access in: 28/05/2025