Development and optimization of a novel in-silico ADMET screening workflow for repurposed FDA-approved drugs targeting drug-resistant bacterial infections

Project Abstract

The rapid emergence of multidrug-resistant bacterial pathogens poses a critical threat to global health, necessitating innovative strategies to identify effective therapeutics with favorable safety profiles. This study presents a novel in-silico ADMET (absorption, distribution, metabolism, excretion, and toxicity) screening workflow designed to repurpose FDA-approved drugs for combating drug-resistant infections, thereby accelerating the drug development pipeline while reducing clinical attrition. The workflow integrates multiple computational modules, including structure-based virtual screening, ligand-based predictive models, physiologically based pharmacokinetic (PBPK) simulations, and mechanistic toxicity assessments, to systematically filter large drug repurposing libraries and prioritize candidates with optimal pharmacokinetic properties and acceptable safety margins. We assembled a curated dataset of FDA-approved compounds with diverse scaffolds and known safety profiles, augmented with pathogen-specific target models derived from resistant bacterial strains. The screening pipeline begins with high-throughput docking against validated bacterial targets, followed by ligand-based similarity analyses to identify potential repurposing candidates with known mechanistic relevance. Predicted ADMET properties are then evaluated using ensemble QSAR models and machine learning classifiers trained on extensive pharmacokinetic and toxicity datasets, enabling robust cross-validation of absorption potential, distribution to infection sites, metabolic stability, and elimination routes. To address the complexities of bacterial infections, we incorporated PBPK simulations to predict tissue concentration-time profiles under different dosing regimens and infection loci, ensuring that effective drug concentrations are achievable in relevant compartments. A novel multi-criteria decision framework was developed to balance efficacy signals against pharmacokinetic feasibility and safety risk, incorporating uncertainty quantification to rank candidates under variable physiological conditions and pathogen burdens. The workflow was validated retrospectively with known successful and failed repurposed antibiotics, demonstrating improved prioritization accuracy compared with conventional single-criterion approaches. Prospective application to a diverse set of FDA-approved compounds yielded several high-potential candidates with plausible mechanisms of action against resistant pathogens, favorable predicted ADMET profiles, and attainable tissue exposures consistent with target inhibition. In-depth case analyses highlight how early integration of ADMET constraints can prevent late-stage attrition and reduce time-to-lead. Sensitivity analyses identified critical parameters driving selection, such as membrane permeability, efflux susceptibility, and hepatotoxicity risk, guiding future optimization strategies. The study also delineates limitations, including gaps in target validation for certain resistant strains and the inherent uncertainty in translating in-silico predictions to in vivo outcomes. Overall, the proposed screening workflow offers a scalable, cost-effective framework for rapid repurposing of approved drugs, with broad applicability across bacterial resistance landscapes and potential adaptability to other infectious disease contexts. The findings underscore the value of integrated ADMET-informed in-silico methodologies to streamline drug discovery and enhance the repertoire of oral, safe, and efficacious treatments for drug-resistant bacterial infections.

Project Overview

What This Project Is About

---

A simple, AI-assisted project exploring how existing FDA-approved drugs can be repurposed to fight drug-resistant bacteria using computer-based screening that predicts how drugs behave in the body and how effective they might be against bacteria.

---

---

The Problem It Addresses

---

Many bacteria are no longer killed effectively by standard medicines. Testing every old and new drug in the lab is slow and expensive. This project aims to speed up the process by using computer models to quickly filter promising candidates for lab tests.

---

---

Objectives of the Project

---

1. Learn how to screen existing drugs for new bacterial targets using in-silico methods.
2. Assess safety and effectiveness predictions to pick strong candidates.
3. Develop a reproducible workflow that others can reuse with different drug sets.
4. Present a ready-to-test list of candidate drugs for experimental verification.

---

---

What You Will Do Step by Step

---

1. Review background on antibiotic resistance and drug repurposing basics.
2. Collect a library of FDA-approved drugs relevant to bacterial targets.
3. Build and validate a computational workflow to predict ADMET (absorption, distribution, metabolism, excretion, toxicity).
4. Run in-silico screens against bacterial targets and rank candidates.
5. Analyze top candidates for potential lab testing and safety concerns.
6. Document methods and create a user-friendly guide for future users.

---

---

Expected Outcome

A concise list of repurposable drugs with predicted safety and efficacy profiles, plus a documented screening workflow that others can apply to different datasets, potentially accelerating experimental validation and guiding future research.