DynaMO, a package identifying transcription factor binding sites in dynamical ChIPSeq/RNASeq datasets, identifies transcription factors driving yeast ultradian and mammalian circadian cycles

Biological processes are usually associated with genome-wide remodeling of transcription driven by transcription factors (TFs). Identifying key TFs and their spatiotemporal binding patterns are indispensable to understanding how dynamic processes are programmed. We present a computational method, dynamic motif occupancy (DynaMO), which exploits random forest modeling and clustering based enrichment analysis. DynaMO exploits TF motifs and dynamic ChIP-seq data of chromatin surrogates such as histone modifications to infer important TFs and their spatiotemporal binding in biological processes. Application of DynaMO to the yeast ultradian cycle, mouse circadian clock and human neural differentiation exhibits its accuracy and versatility. We further demonstrate the function of stress response regulators Msn2 and Msn4 as key TFs activating yeast glycolysis genes. Msn2/4 regulates the cellular transition from quiescence to growth by accumulating the intracellular acetyl-CoA level through glycolysis. We anticipate DynaMO will be generally useful for elucidating transcriptional programs in dynamic processes.生物过程通常与转录因子（transcription factors，TFs）介导的全基因组转录重塑密切相关。鉴定关键转录因子及其时空结合模式，对于解析动态生物过程的编程逻辑不可或缺。本研究提出一种名为动态基序占用（dynamic motif occupancy，DynaMO）的计算方法，该方法整合随机森林建模与基于聚类的富集分析策略。DynaMO利用转录因子基序以及组蛋白修饰等染色质替代标志物的动态染色质免疫共沉淀测序（ChIP-seq）数据，推断生物过程中的关键转录因子及其时空结合特征。将DynaMO应用于酵母超日节律周期、小鼠昼夜节律钟以及人类神经分化数据集，验证了该方法的准确性与通用性。本研究进一步证实，应激反应调控因子Msn2与Msn4作为激活酵母糖酵解基因的关键转录因子发挥功能：Msn2/4通过糖酵解途径提升细胞内乙酰辅酶A水平，从而调控细胞从静息状态向增殖状态的转变。我们预计DynaMO将可广泛用于解析动态生物过程中的转录调控程序。