Deep learning models predict regulatory variants in pancreatic islets and refine type 2 diabetes association signals.

Overall, we observed a modest but highly significant correlation of predictions with both these regulatory prediction scores reported by DeepSEA (multi-tissue: r = 0.227, p<2.2e-16; PanIslets: r = 0.223, p<2.2e-16) ( Figure 3—figure supplement 1 - AB ).

also in 132,142 other papers

Similarly, while the rs17712208 variant at PROX1 reported in this study was found to be regulatory using the multi-tissue mode of DeepSEA (p=6.25e-03), the predictions for pancreatic islets accessibility profiles (PanIslets) were only borderline significant for this variant (p=0.049).

also in 11,409 other papers