Datasets for CRISPR-MBTF Project

This record contains processed, CSV-formatted data from multiple sources, prepared for the CRISPR-MBTF project. Part 1: DeepCRISPR Off-Target Data: The dataset includes off-target data originally published as part of the DeepCRISPR study (Chuai et al., 2018). The original files for the K562 and HEK293T cell lines were converted from their '.epiotrt' format into three user-friendly CSV files: k562_data.csv, hek293t_data.csv, and a combined all_data.csv. These files contain on-target/off-target DNA sequence pairs, four epigenetic features for each site, and a binary label for off-target activity. Part 2: CRISOT Scoring Parameters: The dataset also includes scoring parameters derived from the CRISOT model (Kim et al., 2019), sourced from the bm2-lab/CRISOT GitHub repository. Specifically, the pandas DataFrame object containing model parameters was extracted from the models/crisot_score_param.pkl file and saved as score_param_df.csv. Other Python objects from the original pickle file were not included.Original Sources: 1. Chuai, G., Ma, H., Yan, J. et al. DeepCRISPR: optimized CRISPR guide RNA design for high-fidelity genome editing. Genome Biol 19, 80 (2018). https://doi.org/10.1186/s13059-018-1459-42. Kim, H., Min, S., Song, M. et al. Deep learning-based prediction of CRISPR-Cpf1 guide RNA activity for high-fidelity genome editing. Nat Biotechnol 37, 1317–1326 (2019). https://doi.org/10.1038/s41587-019-0264-6