CRISPR-Cas13系统是CRISPR-Cas基因编辑系统家族中的新成员，其主要功能是对RNA进行定向编辑。CRISPR-Cas基因编辑系统是当前生命科学的突破性重大技术之一，其操作简单，对基因组的编辑效率较高，理论上可对任何物种的基因组进行定向编辑。目前最为常见的是CRISPR-Cas9系统，但该系统主要作用于DNA。故在非破坏性基因治疗以及RNA病毒检测等方面，CRISPR-Cas13系统具有独特的优势。但CRISPR-Cas13系统的实际应用仍遭受很大程度的制约，这其中最重要的因素就是脱靶效应，即向导RNA将编辑系统指向错误的RNA靶区域，导致细胞产生错误的表达谱。因此，本研究将基于当前深度学习领域前沿的基于强化学习方法的自动化深度学习技术，结合迁移学习，建立高精度CRISPR-Cas13脱靶分布预测系统，从而使得高特异性的RNA编辑成为可能。

CRISPR-Cas13 RNA editing system is a new member of the CRISPR-Cas gene editing system family, which can perform guided editing on RNA sequence. CRISPR-Cas gene editing system is currently one of the breakthrough technologies in life science, which is easy to operate ,has a high editing efficiency and theoretically can perform guided editing on genome of any species. CRISPR-Cas9 system is the most commonly used CRISPR-Cas system, but it is mainly applied to DNA. Hence, for non-destructive gene therapy and RNA virus detection, CRISPR-Cas13 system has its unique advantage. However, there are severe issues limiting its application, the most important one of which is off-target effect, namely the guide RNA points the system to the wrong position causing non-targeted region edited, which leading to wrong expression spectrum of the cell. Therefore, based on automated deep learning and transfer learning techniques, this project aims to establishing an off-target distribution prediction system of CRISPR-Cas13 RNA editing system, in order to make high-specificity RNA editing possible.