National Key Laboratory of Crop Genetic Improvement, Center for Bioinformatics, College of Life Science and Technology, Huazhong Agricultural University, Wuhan 430070, China.
Nucleic Acids Res. 2013 Oct;41(19):e183. doi: 10.1093/nar/gkt745. Epub 2013 Aug 21.
The 3D chromatin structure modeling by chromatin interactions derived from Hi-C experiments is significantly challenged by the intrinsic sequencing biases in these experiments. Conventional modeling methods only focus on the bias among different chromatin regions within the same experiment but neglect the bias arising from different experimental sequencing depth. We now show that the regional interaction bias is tightly coupled with the sequencing depth, and we further identify a chromatin structure parameter as the inherent characteristics of Hi-C derived data for chromatin regions. Then we present an approach for chromatin structure prediction capable of relaxing both kinds of sequencing biases by using this identified parameter. This method is validated by intra and inter cell-line comparisons among various chromatin regions for four human cell-lines (K562, GM12878, IMR90 and H1hESC), which shows that the openness of chromatin region is well correlated with chromatin function. This method has been executed by an automatic pipeline (AutoChrom3D) and thus can be conveniently used.
通过 Hi-C 实验得出的染色质相互作用进行 3D 染色质结构建模，受到这些实验中固有测序偏差的显著挑战。传统的建模方法仅关注同一实验中不同染色质区域之间的偏差，但忽略了来自不同实验测序深度的偏差。我们现在表明，区域相互作用偏差与测序深度紧密相关，并且我们进一步确定了一个染色质结构参数，作为染色质区域中 Hi-C 衍生数据的固有特征。然后，我们提出了一种通过使用此识别参数来放松两种测序偏差的染色质结构预测方法。该方法通过对四种人类细胞系（K562、GM12878、IMR90 和 H1hESC）的各种染色质区域进行细胞内和细胞间比较进行了验证，结果表明染色质区域的开放性与染色质功能密切相关。该方法已通过自动流水线（AutoChrom3D）执行，因此可以方便地使用。
