DeepYY1: a deep learning approach to identify YY1-mediated chromatin loops

Area of Science:

• Genomics and Bioinformatics
• Molecular Biology
• Computational Biology

Background:

• The protein Yin Yang 1 (YY1) plays a crucial role in mammalian gene control by mediating interactions between enhancers and promoters.
• Enhancer-promoter interactions are fundamental to gene regulation, often involving chromatin looping.
• Existing computational methods can characterize DNA element interactions but lack specificity for identifying YY1-mediated chromatin loops.

Purpose of the Study:

• To develop a novel computational method for identifying YY1-mediated chromatin loops.
• To investigate the role of DNA sequence in the formation of YY1-mediated chromatin loops.
• To provide a user-friendly tool for researchers studying gene regulation.

Main Methods:

• Development of a deep learning algorithm, DeepYY1, utilizing the word2vec model.
• Training and testing the algorithm on datasets from various cell types to assess prediction performance.
• Analysis of sequence features contributing to YY1-mediated loop formation.

Main Results:

• DeepYY1 achieved high prediction accuracy, with Area Under the Curve (AUC) values greater than or equal to 0.93 on training and testing datasets.
• The model demonstrated excellent performance in identifying YY1-mediated chromatin loops across different cell types.
• Sequence characteristics were identified as significant factors influencing the formation of these chromatin loops.

Conclusions:

• DeepYY1 is an effective deep learning tool for accurately identifying YY1-mediated chromatin loops.
• DNA sequence plays a critical role in the formation of YY1-mediated chromatin loops.
• A publicly accessible web server for DeepYY1 has been established to facilitate research in gene regulation.