Prime editing efficiencies on high-throughput datasets.

We conducted a high-throughput screen in HEK293T to analyze prime editing outcomes of 92,423 pegRNAs on a highly diverse set of 13,349 human pathogenic mutations that include base substitutions, insertions and deletions. Based on this dataset, we identified sequence context features that influence prime editing rates and trained PRIDICT (PRIme editing guide preDICTion), an attention-based bi-directional recurrent neural network. PRIDICT reliably predicts editing efficiencies and unintended editing rates (www.pridict.it). Additionally we provide sequencing datasets of experiments on a smaller library, to validate PRIDICT performance in HEK293T, K562, U2OS and mouse hepatocytes. Sequencing data from arrayed endogenous experiments in HEK293T and K562 is also included.