LinearKnot: linear-time ensemble analysis of RNA density-2 pseudoknots under hierarchical folding

LinearKnot approximates CParty’s ensemble analysis of RNA density-2 pseudoknots with beam pruning, achieving linear-time partition functions and MFE predictions in sequence length.

• Why it matters: Exact hierarchical pseudoknot ensembles are computationally expensive, limiting structure prediction and probability-based analyses when pseudoknots are present.
• What they did: LinearKnot uses beam pruning that combines at most two partial structures per step, producing partition functions, MFE predictions, and representative gamma-centroid structures with loop-profile probabilities and window accessibility.
• The result: With beam width 100, it completed partition functions for all 107 real RNAs up to 29 kilobases; for URS00023A4DB9 (29,106 nt), partition-function, pairing-probability, and MFE took about 7 minutes and 3.22 GiB.

AI summary of the abstract · may contain errors · not medical advice · read the paper.