MLP accompanied beam search for the resonance assignment problem

Resonance signal assignment is a well known problem encountered during the process of biomolecule structure determination based on nuclear magnetic resonance (NMR) spectroscopy. As one of the first steps following the experimental part of the process, it significantly affects the quality of a resulting molecule model. In the paper we present a new approach for automated signal assignment in two‐dimensional NMR spectra of RNA molecules. The proposed algorithm combines beam search with a neural network in order to reconstruct and evaluate solutions. Computational tests with a set of experimental and simulated NMR spectra show an excellent performance of the approach in comparison with the existing methods. The algorithm implementation is freely available at http://www.cs.put.poznan.pl/mszachniuk/bsmlp.zip .