Fig. 2
From: MS2DeepScore: a novel deep learning similarity measure to compare tandem mass spectra
A To better account for the unbalanced nature of the total set of Tanimoto scores we here plotted the MS2DeepScore Tanimoto score predictions on the test set across different bins of Tanimoto scores (< 0.1, 0.1–0.2 etc. until > 0.9). B Average precision measures across the different Tanimoto score bins by RMSE. C The number of spectrum pairs within the test set which fall into each of the 10 Tanimoto score bins, illustrating the highly unbalanced nature of the dataset (see also Fig. 3A for the score distribution across the full dataset)