ORCID
0009-0001-3427-5489
Keywords
Antimicrobial Peptides (AMPs), Generative Adversarial Network (GAN), Graph Neural Network (GNN), Antibacterial Testing, Cell Viability
Subject Categories
Biochemistry | Biomedical Informatics
Abstract
The rapid rise of antibiotic-resistant bacteria has created an urgent need for new antibacterial agents. Antimicrobial peptides (AMPs) are promising candidates because they can kill bacteria through membrane disruption and other cellular mechanisms. In this thesis, an artificial intelligence-guided workflow was used to generate and screen new antimicrobial peptide candidates. A reinforcement learning-based Generative Adversarial Network (GAN) was used to create a large peptide pool. The generated sequences were then filtered using graph neural network (GNN) classifiers. These models were used to predict antimicrobial activity, non-toxicity, bacterial membrane disruption, intracellular mode of action, and antibiofilm activity. An ESM-2 (Evolutionary Scale Modeling) based protein language model classifier was also used to screen for antimicrobial activity. Physicochemical properties, including hydrophobicity and net charge, were also used for final selection. From 118,647 cleaned, generated sequences, 13 peptide candidates were selected for synthesis and experimental testing. The selected peptides were synthesized with high purity and tested for antibacterial activity using minimum inhibitory concentration (MIC) assays. Antibacterial testing was performed against Escherichia coli K-12 and Staphylococcus aureus ATCC 25923 to evaluate activity against Gram-negative and Gram-positive bacteria. The three best peptide candidates were also tested for cytotoxicity using human fibroblast cells. This step was used to check whether the peptides were safe for mammalian cells. Overall, this study combines peptide generation, deep learning-based screening, physicochemical filtering, antibacterial testing, and cell viability testing. This work provides a simple pipeline for early-stage antimicrobial peptide discovery and will support the development of safer, more effective peptide-based antibacterial agents.
Completion Date
2026
Semester
Summer
Committee Chair
Mukhopadhyay, Kausik
Degree
Master of Science in Electrical Engineering (M.S.E.E.)
College
College of Engineering and Computer Science
Department
Electrical and Computer Engineering
Format
Document Type
Thesis
Language
English
Release Date
8-15-2027
STARS Citation
Tapotee, Malisha Islam, "Designing Effective Antimicrobials: A Deep Learning Approach to Antimicrobial Peptide Generation and Screening" (2026). Graduate Studies Theses and Dissertations 2026. 364.
https://stars.library.ucf.edu/gradstudies_etd_2026/364
Accessibility Statement
This item was created or digitized prior to April 24, 2027, or is a reproduction of legacy media created before that date. It is preserved in its original, unmodified state specifically for research, reference, or historical recordkeeping. In accordance with the ADA Title II Final Rule, the University Libraries provides accessible versions of archival materials upon request. To request an accommodation for this item, please submit an accessibility request form.