REFLEX: RNA Ensemble Generation via Flexibility-Calibrated Stochastic Bridge

Key points

• RNA molecules often function through multiple experimentally observed conformational states.1
• A central challenge is to capture ensemble diversity without sacrificing structural fidelity: samples may collapse toward averaged representative structures or drift into unsupported conformational regions.1
• During training, experimental B-factors calibrate residue-wise bridge widths, exposing each observed conformer through local perturbation neighborhoods with residue-specific scales while keeping constrained regions sharper.1
• On a multi-conformer RNA benchmark, REFLEX achieves the best precision-recall trade-off for ensemble coverage across evaluation thresholds, while recovery-stress ablations indicate gains in target-frame recovery under stronger perturbations.1

Sources (1)

• [1]REFLEX: RNA Ensemble Generation via Flexibility-Calibrated Stochastic Bridge
  arXiv (AI, ML, NLP, CV, robotics, multi-agent) · Oct 19, 12:18 PM

  We introduce REFLEX (RNA Ensemble generation calibrated by FLEXibility), a sequence-conditioned generator for RNA backbone-frame ensembles based on a heteroscedastic stochastic bridge on SE(3).

  RNA molecules often function through multiple experimentally observed conformational states.

Extractive summary: sentences quoted from the sources.