Computer-assisted molecular design
Our research mainly focuses on leveraging computer-assisted molecular modeling to design and discover new therapeutic agents. Our expertise spans protein-ligand interaction modeling, structure- and ligand-based drug design, molecular dynamics with enhanced sampling techniques, QSAR/QSPR analyses, and the integration of machine learning in drug discovery pipelines.
Polypharmacology
Our lab is involved in several drug discovery campaigns focused on identifying multi-target ligands. These efforts target widely researched pharmacological proteins, such as kinases and GPCRs, alongside disease-specific proteins. Our in silico techniques include classical molecular docking, precision-tuned library generation, and AI-driven data analysis. Recent successes include the development of a dual GSK-3 beta/D3 receptor ligand for bipolar disorder treatment and a multitarget-directed ligand for multiple sclerosis with combined anti-inflammatory and remyelinating properties (MARs).
Ligand binding kinetics and free-energy calculations
Our group is proficient in the use of different enhanced sampling techniques, including metadynamics, adiabatic-bias MD, umbrella sampling and accelerated MD, for the prediction of binding free-energy and kinetics of systems of pharmaceutical interest. These approaches are exploited for the study of ligand binding and unbinding events, conformational transitions in protein-ligand systems and to develop post-processing workflows for virtual screening results.
Development of a platform for efficient preclinical drug discovery
We contribute to developing and maintaining an integrated platform that combines a HPC infrastructure with biophysical assays (SPR, MST, and CD) to support both academic and industrial drug discovery efforts, accessible to third-party collaborators. Our platform enables virtual screening of ultra-large libraries using structure- and ligand-based methods, incorporates machine learning for compound prioritization, and supports multi-scale workflows for identifying covalent ligands. Additionally, our screening protocols can employ a proprietary virtual library featuring compounds synthesized at the University of Urbino.
Funded by PNRR – Missione 4 – Componente 2 – Progetto “Innovation, digitalisation and sustainability for the diffused economy in Central Italy” (VITALITY)
GPCRs allostery, including ALLODD – Allostery in Drug Discovery
This research project focuses on developing computational methods to enhance our understanding of allostery in G-protein coupled receptors (GPCRs), with particular emphasis on the protein-membrane interface.
Activities
- Developing a binding-site detection algorithm that leverages cholesterol to identify and characterize potential binding hotspots.
- Creating a structural cheminformatics platform that integrates chemical, biological, and structural data on GPCR allosteric interactions. This platform will support the identification of structural determinants for allosteric modulation, enabling structure-based ligand design and advancing computer-aided drug discovery methods for virtual screening and the structure-based design of allosteric modulators.
- Combining classical molecular dynamics simulations with pocket-pocket crosstalk analysis to assess protein communication pathways.”
Funded by EU Horizon 2020 MSCA Program under grant agreement 956314 (ALLODD)
Molecular properties prediction with Deep-Learning algorithm.
Our team has recently used deep learning for predicting molecular properties. We primarily focus on environmental toxicity predicting Persistence, Bioaccumulation, and Toxicity (PBT) profiles of biologically active compounds and docking-based molecular prioritization within ultra-large libraries.