Scholarlink

CRISPR-Cas9 technology, in spite of a huge progress it has undergone since its discovery, is still hindered by a phenomenon of off-targets occurrence. Experimental techniques of their detection are expensive and yet not fully credible thus encouraging the pursuit for effective in silico methods mimicking the functionality of the experimental ones. In this paper we shift our attention from the issue of the quality of particular machine learning and deep learning models described in the literature to answering the question if existing evaluation methods and measures are sufficient to make credible judgements on the model’s quality. Lastly, we provide some directions and hints pertaining to the evaluation of the models dealing with the issue of off-targets prediction.

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

Contextual Statistical Evaluation of Selected CRISPR-Cas9 Recurrent Deep Learning Models Predicting Off-Target Activities for K562 and Hek293 Cell Lines

  • 会议论文

  • 发表日期：

    2025年

查看全文
  • Maciej Powierża,
  • Łukasz Łaczmański,
  • Maciej Huk

摘要

CRISPR-Cas9 technology, in spite of a huge progress it has undergone since its discovery, is still hindered by a phenomenon of off-targets occurrence. Experimental techniques of their detection are expensive and yet not fully credible thus encouraging the pursuit for effective in silico methods mimicking the functionality of the experimental ones. In this paper we shift our attention from the issue of the quality of particular machine learning and deep learning models described in the literature to answering the question if existing evaluation methods and measures are sufficient to make credible judgements on the model’s quality. Lastly, we provide some directions and hints pertaining to the evaluation of the models dealing with the issue of off-targets prediction.