Mountain Centroid: RNA Ensemble Representation with Mountain Profiles
Background: RNA molecules form thermodynamic ensembles, but interpretation often requires a single representative structure. Existing base-pair centroid estimators assess agreement at the level of individual base pairs and do not directly target nesting depth along the sequence. Methods: We introduce Mountain Centroid, which minimizes expected squared mountain-profile distance, and derive dynamic…
RNA molecules often form multiple thermodynamic ensembles, necessitating a single representative structure for interpretation. Current base-pair centroid estimators evaluate agreement at the level of individual base pairs, disregarding nesting depth along the sequence. The authors present Mountain Centroid, which minimizes the expected squared mountain-profile distance and derives dynamic programming algorithms with and without RNA pairing constraints. They also integrate Mountain Centroid with the base-pair centroid gain.
The Mountain Centroid algorithm was tested against 21,254 RNAStrAlign sequences. It demonstrated a lower median normalized mean squared mountain distance (NMSMD) compared to minimum-free-energy (MFE) and base-pair centroid structures with a weight (γ) of 1. However, its median base-pair F1 score was lower. Implementing RNA pairing constraints improved base-pair F1 for 59.35% of sequences while reducing it for 3.58%.
At a specific weight, the combined objective structure exhibited a median base-pair F1 score comparable to MFE while maintaining a lower median NMSMD than MFE and all tested γ-centroid settings.
In conclusion, Mountain Centroid offers a novel way to represent an RNA structural ensemble as a single secondary structure. By combining mountain-profile and individual-base-pair criteria, researchers can adjust the relative contributions of these factors to suit their needs.
Written by urgent.news from bioRxiv's reporting — not their text. Machine-written — may contain errors; check the original before relying on it.
