Scholarlink

The precise design of single-guide RNA (sgRNA) remains a central requirement for reliable CRISPR-Cas9 gene knockout experiments. We present a command-line module that automates sgRNA design by combining the following: (i) programmatic retrieval of target gene sequences from Ensembl, (ii) exhaustive NGG PAM scanning on both strands, (iii) 30-bp context construction, (iv) prediction of on-target efficiency via the DeepSpCas9 deep learning model, and (v) whole-genome off-target search against GRCh38 using Bowtie, with concise reporting for candidates having less than 5 predicted off-target loci. The tool also supports exon-only or intron-only search modes to reflect experimental constraints. In tests on TP53, BRCA1, and CD36, the pipeline generated approximately 14,000 candidates; top-ranked sgRNAs for TP53 achieved high predicted efficiency (0.845) with only one predicted off-target, and exon-localized guides showed higher average efficiency than intronic ones. The module prioritizes interpretability, speed less than 3 min for typical genes, and extensibility while maintaining accuracy through machine learning based scoring and genome-scale specificity checks.

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

A Computational Module for sgRNA Design and Analysis Using the CRISPR-Cas9 Knock-Out System

  • 会议论文

  • 发表日期：

    2026年

查看全文
  • Georgiana Nicoleta Stan,
  • Gabriela Niculescu

摘要

The precise design of single-guide RNA (sgRNA) remains a central requirement for reliable CRISPR-Cas9 gene knockout experiments. We present a command-line module that automates sgRNA design by combining the following: (i) programmatic retrieval of target gene sequences from Ensembl, (ii) exhaustive NGG PAM scanning on both strands, (iii) 30-bp context construction, (iv) prediction of on-target efficiency via the DeepSpCas9 deep learning model, and (v) whole-genome off-target search against GRCh38 using Bowtie, with concise reporting for candidates having less than 5 predicted off-target loci. The tool also supports exon-only or intron-only search modes to reflect experimental constraints. In tests on TP53, BRCA1, and CD36, the pipeline generated approximately 14,000 candidates; top-ranked sgRNAs for TP53 achieved high predicted efficiency (0.845) with only one predicted off-target, and exon-localized guides showed higher average efficiency than intronic ones. The module prioritizes interpretability, speed less than 3 min for typical genes, and extensibility while maintaining accuracy through machine learning based scoring and genome-scale specificity checks.