Tailor‐made computational protocols for precise characterization of small biological building blocks using QM and MM approaches

Abstract

Computational modeling involving Quantum Mechanics (QM) and Molecular Mechanics (MM) calculations are widely utilized to unveil the atomic-molecular properties that underpin their inherent characteristic features. The choice over the either of the QM and MM methods or a multiscale composite approach is driven by the target property of interest, and of course, the molecular size. Often, tailor-made schemes need to be devised to match the specific study purpose. Herein, we provide a perspective of these approaches addressing their effectiveness in terms of the delicate balance between the accuracy and computational feasibility. We focus on representative examples to highlight how different approaches can be fruitfully exploited for modeling the conformational landscape, and possibly, the spectroscopic behavior of biochemical molecules, especially amino acids building blocks.

Accuracy and Interpretability: The Devil and the Holy Grail. New Routes across Old Boundaries in Computational Spectroscopy

“…Rotational spectroscopy is the technique of choice when aiming at retrieving accurate geometries of molecules, but it also provides insights into bond topology and non-covalent interactions, as well as into internal dynamics. Hence, it allows detailed investigations of biochemical building blocks and molecular complexes, showing the prototypical interactions ruling the behavior and functionality of biomolecules without the concerns due to competing environmental effects (see refs and , and the cited literature). In addition, rotational spectroscopy is particularly well suited for the unequivocal detection of chemical species in the gas phase so that, over the last decades, it has become the preferred method for astronomical detections. , However, the analysis of rotational spectra is often far from being a simple task, in particular when the size of system grows, thus leading to the presence of complicated conformational/reactive mixtures, or when hyperfine interactions come into play because of the presence, for example, of quadrupolar nuclei or unpaired electrons.…”

Computational Spectroscopy in Solution by Integration of Variational and Perturbative Approaches on Top of Clusterized Molecular Dynamics

“…Since the accuracy of the classical sampling strictly depends on the quality of the force-field (FF) parameters, the availability of an accurate FF is the mandatory first step of any successful modeling. 10,12,13 Once an FF is available, molecular dynamics (MD) simulations can be used to sample the phase space usually employing periodic boundary conditions (PBC). 14 Unfortunately, PBC are not free from possible artifacts for intrinsically nonperiodic systems, 15 and therefore, several alternative strategies enforcing nonperiodic boundary conditions (NPBC) have been proposed.…”

“…The most effective solution to this problem is offered by multiscale strategies like quantum mechanics–molecular mechanics (QM/MM) approaches in which a relatively small part of the system (e.g., the chromophore) is treated at the highest possible QM level, whereas the remaining (huge) part (including remote regions of the solute and the solvent possibly beyond the cybotactic zone) is treated at a lower QM or MM level. − …”

“…The description of a molecular system at the MM level requires a set of parameters encoding its properties (force field (FF)). Since the accuracy of the classical sampling strictly depends on the quality of the force-field (FF) parameters, the availability of an accurate FF is the mandatory first step of any successful modeling. ,, …”

“…The simplest way to perform this subsampling is to extract snapshots from the MM trajectories with a constant step, but this strategy is both inefficient and scarcely insightful. Unsupervised learning (UL) techniques such as clustering, ,, self-organizing maps, and combinatorial optimization may yield a balanced and efficient subsampling of MM trajectories once an exhaustive overall sampling has been carried out. The application of UL requires the choice not only of an efficient sampling/classification method but also of suitable molecular descriptors for the comparison of structures.…”