Maurizio Sironi

Accurate simulation of protein–ligand interactions remains a central challenge in drug discovery. Coarse-grained (CG) molecular dynamics offers a route to broader system coverage and efficient sampling by reducing molecular resolution. Here, we examine the potential of Martini 3 for drug discovery, reviewing methodological advances in simulation setup and addressing persistent challenges in sampling, model parameterization, and the reliable estimation of binding thermodynamics. Finally, we discuss priorities for translating CG simulations into robust, scalable drug discovery workflows.

Accessing deeply buried binding sites remains a major challenge in structure-based drug discovery, where accurate description of both protein dynamics and ligand binding pathways is required. Funnel metadynamics enables simulation of complete binding processes but is computationally demanding at the all-atom resolution. By adopting the Martini 3 force field, coarse-grained funnel metadynamics (CG-FMD) substantially reduces computational requirements while retaining enhanced sampling capabilities. In this work, we assess the capability of CG-FMD to model ligand recognition at the deeply buried colchicinoids site of the tubulin αβ -heterodimer, a multisite protein of strategic importance. We investigated the binding of colchicine, podophyllotoxin and combretastatin-A4, recovering free energy profiles with improved statistical convergence compared to AA-FMD and comparable to experimental references. In particular CG-FMD binding free energies present mean absolute errors between 3 and 10 kJ mol ^{− 1} . These results propose CG-FMD as an efficient, physics-based framework for probing ligand binding to challenging sites.

Despite considerable advances in computational chemistry, bridging the gap between the accuracy of all-atom molecular dynamics (AA-MD) and the high-throughput capabilities of docking remains an unsolved problem in protein-ligand binding free energy predictions. In this work, we propose to address this challenge through coarse-grained funnel metadynamics (CG-FMD) with the Martini 3 force field. This approach combines the reduced computational cost of a CG representation with state-of-the-art enhanced sampling techniques and the interpretability of a physics-based force field. Specifically, the binding of colchicine to two different protein targets was modeled at both AA and CG resolutions, and the corresponding ΔG_bind predictions were compared with experimental references. Additionally, the robustness of CG-FMD-based ΔG_bind predictions was evaluated with respect to various aspects of the simulation setup by collecting more than 7 ms of CG-FMD simulations. The optimal simulation protocol has been further validated against a limited set of compounds chemically different from colchicine. The results demonstrate that CG-FMD yields ΔG_bind estimates comparable to experimental values while requiring only a fraction of the computational cost of AA-MD simulations. Moreover, the extensive sampling achievable with CG-FMD reduces statistical uncertainty in the final predictions, effectively compensating for the simplified system representation. Future work should build upon these methodological insights to broaden the scope of ligands and targets explored.

Despite considerable advances in computational chemistry, bridging the gap between the accuracy of all-atom molecular dynamics (AA-MD) and the high-throughput capabilities of docking remains an unsolved problem in protein-ligand binding free energy predictions. In this work, we propose to address this challenge through coarse-grained funnel metadynamics (CG-FMD) with the Martini 3 forcefield. This approach combines the reduced computational cost of a CG representation with state-of-the-art enhanced sampling techniques and the interpretability of a physics-based forcefield. Specifically, the binding of colchicine to two different protein targets was modeled at both AA and CG resolutions, and the corresponding ΔGbind predictions were compared with experimental references. Additionally, the robustness of CG-FMD-based ΔGbind predictions was evaluated with respect to various aspects of the simulation setup by collecting more than 7 milliseconds of CG-FMD simulations. The results demonstrate that CGFMD yields ΔGbind estimates comparable to experimental values while requiring only a fraction of the computational cost of AA-MD simulations. Moreover, the extensive sampling achievable with CG-FMD reduces statistical uncertainty in the final predictions, effectively compensating for the simplified system representation. Future work should build upon these methodological insights to broaden the scope of ligands and targets explored.

A detailed study of the X...N (X = I, Br) halogen bonds in complexes formed by an extended set of substituted pyridines with D—X molecules (D = X, CN) is reported here. The nature of these interactions has been investigated at different (MP2 and DFT) levels of theory through Bader's quantum theory of atoms in molecules (QTAIM) and Pendás' interacting quantum atoms (IQA) scheme, focusing on the role of the local environment (i.e. the substituent on the pyridine ring and the halogenated residue) on the halogen bond features. We found that the exchange-correlation energy represents a substantial contribution to the IQA total energy, in some cases comparable to (I₂ complexes) or even dominating (ICN complexes) the electrostatic term. Meaningful information is provided by the source function, indicating that the major contribution to the electron density at the bond critical point of the X...N interaction is derived from the halogen atom, while a much lower contribution comes from the nitrogen atom, which acts as either source or sink for electron density. A relevant contribution from distal atoms, including the various electron-donor and electron-withdrawing substituents in different positions of the pyridine ring, is also determined, highlighting the non-local character of the electron density. The existence of possible relationships between binding energies, interaction energies according to IQA, and QTAIM descriptors such as delocalization indices and source function, has been inspected. In general, good correlations are only found when the local environment, external to the directly involved halogen and nitrogen atoms, plays a minor role in the interaction.

Microanalysis techniques based on resonance Raman and reflection visible spectroscopy have been applied to the characterization of pigments responsible for the blue or violet coloration in flowers; in particular of Lobelia erinus, Campanula portenschlagiana, Cineraria, Viola tricolor, Anemone coronaria, Agapanthus, Platycodon, Salvia farinacea, Plumbago capensis, Ceratostigma plumbaginoides, Commelina communis and Salvia patens. The spectroscopic methods were applied both in vivo on the flower petals and in vitro on extracts obtained through a procedure based on SPE (solid-phase extraction) optimized for minimal quantities of vegetable raw material. Different patterns obtained for the Raman spectra have been correlated, also on the basis of density functional theory (DFT) calculations, with different schemes of substitution of the benzopyrilium nucleus of the anthocyanins and with various possible forms of copigmentation responsible for the stabilization of the blue color. The results obtained were verified by comparison with the analysis of the extracts by HPLC-ESI-MS (liquid chromatography-electrospray ionization-mass spectrometry).

Molecular similarity is an impressively broad topic with many implications in several areas of chemistry. Its roots lie in the paradigm that ‘similar molecules have similar properties’. For this reason, methods for determining molecular similarity find wide application in pharmaceutical companies, e.g., in the context of structure-activity relationships. The similarity evaluation is also used in the field of chemical legislation, specifically in the procedure to judge if a new molecule can obtain the status of orphan drug with the consequent financial benefits. For this procedure, the European Medicines Agency uses experts’ judgments. It is clear that the perception of the similarity depends on the observer, so the development of models to reproduce the human perception is useful. In this paper, we built models using both 2D fingerprints and 3D descriptors, i.e., molecular shape and pharmacophore descriptors. The proposed models were also evaluated by constructing a dataset of pairs of molecules which was submitted to a group of experts for the similarity judgment. The proposed machine-learning models can be useful to reduce or assist human efforts in future evaluations. For this reason, the new molecules dataset and an online tool for molecular similarity estimation have been made freely available.

The concept of orthogonality between halogen and hydrogen bonding, brought out by Ho and coworkers some years ago, has become a widely accepted idea within the chemists’ community. While the original work was based on a common carbonyl oxygen as acceptor for both interactions, we explore here, by means of M06-2X, M11, ωB97X, and ωB97XD/aug-cc-PVTZ DFT calculations, the interdependence of halogen and hydrogen bonding with a shared π-electron system of benzene. The donor groups (specifically NCBr and H₂O) were placed on either or the same side of the ring, according to a double T-shaped or a perpendicular geometry, respectively. The results demonstrate that the two interactions with benzene are not strictly independent on each other, therefore outlining that the orthogonality between halogen and hydrogen bonding, intended as energetical independence between the two interactions, should be carefully evaluated according to the specific acceptor group.