FIGURE 3.

Comparison of performance for replicates trained with the best parameter combinations for each input type. (A). Boxplot comparing the perfect match fraction found when each model replicate was run against the test set. (B). Boxplot showing the modal offset of each prediction across all model replicates. (C). Histogram showing the frequency of positional shifts of predicted cut sites relative to their annotated locations for the best replicate trained using nucleotide, RNAfold, and bpRNA structure array. D. A line plot showing the fraction of cut sites identified within varying distances from the annotations for the best replicate trained on nucleotide, RNAfold, and bpRNA structure array.