Extended Data Fig. 5: Rejected best-fit machine learning models for enhancers NEU1 and NEU2.
From: In vivo mapping of mutagenesis sensitivity of human enhancers
(A) Correlations between model predictions and in vivo results. Dots = mutagenized constructs. Black fit line is linear regression. R2 is Spearman correlation. N = 35, 45 (NEU2, NEU1). Asterisk = non-significant (FDR > 0.01). See Supplementary Data 2 for source data. (B) Final TF binding motif and activity map including verified binding motifs discovered through in silico saturation mutagenesis (NEU2 PITX/OTX2 site marked with asterisks). (C) Predicted motif disruptions. Note that validation of the GATA motif in blocks 19–20 did not succeed. (D) Discovery and validation of an additional PITX/OTX2 site in enhancer NEU2.
