UnidecNMR: a new method for automatic NMR peak detection in 1–4 dimensions

The Baldwin group at Oxford Chemistry has introduced a new computational approach for picking peaks in spectra, recently published in Nature Communications.

The proposed strategy exploits the unidec Bayesian deconvolution algorithm that they previously derived, which were originally intended for analysing native mass spectrometry data. This algorithm can remove the effect of a point spread function from any spectral data.

The team’s algorithm can be embedded into a wide range of work flows. It runs quickly on 1–4D NMR data, and can be downloaded from unidecNMR.chem.ox.ac.uk. The group originally used this technique in one dimension as part of a study into pathogen-sugar interactions published in Science.

The algorithm is benchmarked against synthetic and ‘real’ experimental data and outperforms all existing approaches, including those that use artificial intelligence. The group remain convinced that while artificial intelligence can do amazing things, human intelligence and algorithm design is alive and well.

Read more in Nature Communications.