Maximally informative low-resolution representations of proteins

highlight key intermediate steps along the folding pathway

Roberto Menichetti *1,2

1Physics Department, University of Trento, Via Sommarive 14, Trento, Italy.

2Trento Institute for Fundamental Physics and Applications - INFN TIFPA, Via Sommarive 14,

Trento, Italy.

The main challenge of an in silico investigation of biological systems is nowadays shifting from

the production of data to the development of techniques enabling their simple, case-agnostic

interpretation. In this endeavor, high-complexity simulation datasets are often projected onto

a subset of the system's original degrees of freedom to distill biologically meaningful insight.

This projection, however, is frequently enforced a priori based on physicochemical intuition,

and if not chosen properly it risks hampering the signal-to-noise discrimination [1].

The mapping entropy optimization workow (MEOW) is a recently developed method that

aims at addressing this issue [1,2]. Rather than enforcing them, MEOW identies the opti-

mal reduced representations for the system at hand under the requirement of preserving the

maximum amount of statistical information on the original, high-resolution reference. When

applied to all-atom simulations of several proteins, MEOW was found to single out, in an

unsupervised manner, regions of particular functional relevance of these systems, such as sites

involved in substrate binding or catalysis [2-4].

It is natural to expect, however, the importance of a region in a specic process to vary as

the system performs its function, e.g., throughout a conformational change. In this work, we

thus combine MEOW and transition path theory [5] in a unied framework, and apply this

scheme to detect the relevant sites of the chignoling miniprotein at each stage of its folding

transition. The biological insight provided by the protocol on this simple case paves the way

for its successful application to more complex macromolecules.

[1] M. Giulini, M. Rigoli et al., Frontiers in Molecular Biosciences 8, 676976 (2021).

[2] M. Giulini, R. Menichetti, M. S. Shell, and R. Potestio, Journal of Chemical Theory and

Computation 16, 6795 (2020).

[3] M. Giulini, R. Fiorentini, L. Tubiana, R. Potestio, and R. Menichetti, Journal of Chemical

Information and Modeling 64, 4912 (2024).

[4] M. Rigoli, R. Potestio, and R. Menichetti, The Journal of Physical Chemistry B 129, 611

(2025).

[5] E. Vanden-Eijnden, Journal of Statistical Physics 123, 503 (2006).