Table 1.

Settings of the MD simulations of the datasets used in the work

Molecule	Basis set	Temp.	Thermostat	Coef.
Ac-Ala3-NHMe	tight	500	Global Langevin	2
AT-AT	tight	500	Global Langevin	2
Buckyball catcher	light	400	Nosé-Hoover	1700
Lactose	light	500	Nosé-Hoover	1700
Palmitic acid	light	500	Nosé-Hoover	1700

Temperature is given in K. PBE stands for the Perdew-Burke-Ernzerhof functional and MBD stands for many-body dispersion⁷⁴. Coefficient refers to the friction coefficient (in fs) for the global Langevin thermostat, and to the effective mass (in cm⁻¹) for the Nosé-Hoover thermostat.