Fig. 3
From: ReorientExpress: reference-free orientation of nanopore cDNA reads with deep learning
Read types and sequence motifs. The proportion of cDNA reads that were unambiguously mapped to each transcript type (y axis) and classified as correct (True) or incorrect (False) (x axis) by the CNN model (a) and the MLP model (b). All transcript type annotations from the autosomes and sex chromosomes with more than 10 reads mapped are represented in the plot. c The motifs derived from three CNN filters with significant matches to previously described RNA-binding motifs (Additional file 3): filter M24 (RBM42, q-value = 0.0165997), filter M17 (HuR, q-value = 0.0312701), and filter M9 (PCBP1, q-value = 0.0426708). As information content (y axis) is low, the axis scale is shown between 0 and 1