Semiclassical spectroscopy is a practical way to get an accurately approximate quantum
description of spectral features starting from
ab initio
molecular
dynamics simulations. The computational bottleneck for the method is represented by the
cost of
ab initio
potential, gradient, and Hessian matrix estimates.
This drawback is particularly severe for biological systems due to their unique
complexity and large dimensionality. The main goal of this manuscript is to demonstrate
that quantum dynamics and spectroscopy, at the level of semiclassical approximation, are
doable even for sizable biological systems. To this end, we investigate the possibility
of performing semiclassical spectroscopy simulations when
ab initio
calculations are replaced by computationally cheaper force field evaluations. Both
polarizable (AMOEBABIO18) and nonpolarizable (AMBER14SB) force fields are tested.
Calculations of some particular vibrational frequencies of four nucleosides, i.e.,
uridine, thymidine, deoxyguanosine, and adenosine, show that
ab initio
simulations are accurate and widely applicable. Conversely, simulations based on
AMBER14SB are limited to harmonic approximations, but those relying on AMOEBABIO18 yield
acceptable semiclassical values if the investigated conformation has been included in
the force field parametrization. The main conclusion is that AMOEBABIO18 may provide a
viable route to assist semiclassical spectroscopy in the study of large biological
molecules for which an
ab initio
approach is not computationally
affordable.