Skip to main content
Fig. 4 | Genome Biology

Fig. 4

From: Enhanced Integrated Gradients: improving interpretability of deep learning models using splicing codes as a case study

Fig. 4

Identification of A1CF as a regulator of the liver splicing program. a Tomtom [28] alignment of PSSMs for the in vitro determined binding site of the RNA binding protein A1CF ([29] top) and a significant meta-feature identified in multiple liver versus other tissue comparisons (bottom). b Boxplots showing the expression level of A1cf in transcripts per million (TPM) in the indicated tissues across the six mouse replicates. c Motif maps showing the frequency of 3-mers known to bind A1CF (AAU, UAA, or AUU [29, 34]) around the 3’ and 5’ splice sites of cassette exon sets indicated in the legends. Frequencies were smoothed using a running mean of 20 nucleotides (nts). Grayscale boxes indicate significant differences in motif occurrence (− log10(p)) between the regulated versus non-regulated exon sets (p<0.05, Fisher’s exact test assessed at sliding windows of 20 nts). d Bar chart showing the fraction of exons in the indicated sets that contained evidence of A1CF binding proximal to the cassette exon (within 300 nucleotides upstream, within the cassette exon, and/or within 300 nucleotides downstream). Significance was assessed using a two-tailed Fisher’s exact test. e Venn diagram showing the overlap of cassette exons that were alternatively spliced in the liver versus other tissues and those regulated by A1CF, given they were quantified in both experiments. Significance was assessed using a two-tailed Fisher’s exact test

Back to article page