Berlin 2024 – scientific programme

Parts | Days | Selection | Search | Updates | Downloads | Help

BP: Fachverband Biologische Physik

BP 1: Systems and Network Biophysics

BP 1.7: Talk

Monday, March 18, 2024, 11:30–11:45, H 0112

Analysing biological systems via maximally informative representations — •Roberto Menichetti^1,2, Riccardo Aldrigo¹, and Raffaello Potestio^1,2 — ¹Physics Department, University of Trento, Trento, Italy — ²INFN-TIFPA - Trento Institute for Fundamental Physics and Applications, Trento, Italy

The main challenge of an in silico investigation of biological systems is nowadays shifting from the generation of data to the problem of developing techniques enabling their rational interpretation. Often, the noise/signal discrimination passes through dimensionality reduction strategies; while such coarsening is necessary to interpret high-dimensional simulation datasets, it inevitably results in a loss of information on the system that critically depends on the choice of the low-dimensional projection [1]. We here discuss a recent workflow aimed at identifying simplified representations of a system that retain the largest amount of information on its statistical properties while reducing the observational level of detail [1,2]. The protocol is applied to proteins and memory-retrieving neural networks; in both cases, the resulting representations are shown to single out biologically relevant regions of the system, either in the form of functional chemical fragments in the analysed proteins or of strongly coupled neurons in the network. Our results suggest that this scheme can be employed to extract insight from large simulation datasets, further shedding light on the relation between dimensionality reduction and information loss. [1] Giulini M. et al., Front. Mol. Biosci. 8, 676976 (2021). [2] Giulini M. et al., J. Chem. Theory Comput. 16, 6795 (2020).

Keywords: Coarse-graining; Information theory; Biomolecules; Neural networks