Year: 2024
Type: preprint
Abstract: Lateral nano- or microdomains are a major hallmark of biological membranes. A key challenge in analyzing membrane domains from molecular dynamics (MD) simulations lies in accurately assigning lipids to more ordered (Lo) or disordered (Ld) membrane domains, particularly when these domains exhibit similar chemical compositions but differ in structural characteristics. To address this issue, we present the implementation of DomHMM, a novel application previously developed for the unsupervised classification of Lo and Ld domains based on coarse-grained or all-atom MD simulations. DomHMM employs a Gaussian-based Hidden MarkovModel (HMM) for initial lipid state assignment based on specific lipid characteristics, followed by state reweighing through local spatial correlations andsubsequent clustering. This approach significantly improves the efficiency and accuracy of membrane domain analysis, establishing DomHMM as a robust tool for investigating membrane domain behavior.
Source: Research Square
Authors: Eren Tunç, Marius Trollmann, Rainer Böckmann
Institution: Friedrich-Alexander-Universität Erlangen-Nürnberg
Language: English
Subfield: Molecular Biology
Domain: Life Sciences
Open Access status: green
OpenAlex ID: W4402977458
