AbstractThe identification of the protospacer adjacent motif (PAM) sequences of Cas9 nucleases is crucial for their exploitation in genome editing. Here we develop a computational pipeline that was used to interrogate a massively expanded dataset of metagenome and virome assemblies for accurate and comprehensive PAM predictions. This procedure allows the identification and isolation of sequence-tailored Cas9 nucleases by using the target sequence as bait. As proof of concept, starting from the disease-causing mutation P23H in the RHO gene, we find, isolate and experimentally validate a Cas9 which uses the mutated sequence as PAM. Our PAM prediction pipeline will be instrumental to generate a Cas9 nuclease repertoire responding to any PAM requirement.