DR-160 Predicting the target-landscape of kinase-inhibitors with explainable 3D convolutional neural networks enables selection of highly personalized therapies

Many tumors show mutations of kinase proteins. When these kinase proteins are mutated, they can provide a catastrophic effect because this favors cell division over cell death of tumor cells. Since every patient can have multiple kinase mutations, this provides options to inhibit these kinases simultaneously. Since kinase inhibitors commonly inhibit multiple kinases at once, their multiple targets could be optimally matched to individual patients. By using laboratory experiments, we have found that simultaneous inhibition of multiple mutated kinases can lead to more effective therapies. Here we aim to provide a proof of concept that optimally matching drug-targets to patients with multiple kinase mutations can lead to more effective therapies.

Bart Westerman VU Medical Center, Amsterdam UMC the Netherlands