Fig. 1.

t-SNE projection with the well-known pre-miRNAs (in orange) and a random set of unlabeled sequences (in blue) for each genome. Sequences that are closer in the projected feature space have more similar features.