Large DNA loops encompassing genes and their regulatory elements depend on the chromatin architectural protein CCCTC-binding factors (CTCF) interactions. However, most enhancer-promoter interactions do not employ structural protein CTCF. The protein Yin Yang 1 (YY1) has been reported to bind hypo-methylated DNA sequences form homodimers, and can contribute to enhancer-promoter interactions in a manner analogous to DNA looping mediated by CTCF (Fig A).
    we developed a deep learning algorithm named DeepYY1 based on word2vec to determine whether a pair of YY1 motifs would form a loop. The proposed models showed high prediction performance (AUC ≥ 0.93) on both training datasets and testing datasets in different cell types, demonstrating that DeepYY1 has excellent performance in the identification of YY1-mediated chromatin loops (Fig B).