2026-10-12T06:04:05Z https://www.scidb.cn/oai

10.57760/sciencedb.011xb 2026-09-23T19:49:14Z

2026-09-23 Escherichia coli CRISPR-Cas9 sgRNA On-target Activity Dataset (Wild-type, eSpCas9, and recA-Knockout Backgrounds) doi:10.57760/sciencedb.011xb en This dataset is the supporting data for the paper "Prediction Model for Prokaryotic sgRNA Cleavage Activity Fusing Local Multi-Scale Convolution and Global RNA Semantics" (submitted to《Progress in Biochemistry and Biophysics》, manuscript No.20260295). The data were mainly derived from the construction of a prokaryotic sgRNA cleavage-activity prediction model and related supplementary experiments conducted at the College of Science, Dalian Maritime University, from December 2025 to September 2026. The dataset integrates three prokaryotic benchmark datasets—wild-type Cas9 and its high-fidelity variants eSpCas9 and knoRecA_Cas9—sourced from an E. coli genome-scale sgRNA library (Wang et al., Nucleic Acids Research, 2018), comprising more than 137,000 sgRNA sequences of 43 nt together with their cleavage activities measured by high-throughput sequencing, and balanced binary labels (high-/low-efficiency) generated using the median editing efficiency of each dataset as the threshold. It also records the 640-dimensional global semantic features extracted by the pretrained RNA foundation model RNA-FM, the source code and trained weights of the dual-channel fusion model (local multi-scale convolution + global RNA semantics), and the results of supplementary experiments including cross-species few-shot fine-tuning, label-noise robustness, and dual-channel fusion-weight attribution. This dataset lays the foundation for the precise preliminary screening of high-activity sgRNAs and the development of new methods for sgRNA cleavage-activity prediction in prokaryotic CRISPR-Cas9 genome editing. sgRNA; CRISPR-Cas9; cleavage activity prediction yao yue PUBLIC https://creativecommons.org/licenses/by/4.0/ dataset Science Data Bank