Fig. 1.

Texting mining algorithm to search for targetable rare diseases. We filtered all Mendelian diseases with known mutated genes in the OMIM for gain-of-function and late clinical onset related terms for the fist step of filtering to determine our disease targets. In a parallel branch of the pipeline, we filtered for diseases due to loss-of-function mutations with known allosteric activators. All candidates must have a solved protein structure. We manually verified the final disease list to ensure each disease mechanism and onset match our computationally generated label