Non-coding regions of the human genome are important for functional regulations, but their mechanisms remain elusive. We used machine learning to guide a CRISPR screening on hubs (i.e. non-coding loci forming many 3D contacts) and significantly increased the discovery rate of hubs essential for cell growth. We found no clear genetic or epigenetic differences between essential and nonessential hubs, but we observed that some neighboring hubs in the linear genome have distinct spatial contacts and opposite effects on cell growth. One such pair in an epigenetically quiescent region showed different impacts on gene expression, chromatin accessibility and chromatin organization. We also found that deleting the essential hub altered the genetic network activity and increased the entropy of chromatin accessibility, more severe than that caused by deletion of the nonessential hub, suggesting that they are critical for maintaining an ordered chromatin structure. Our study reveals new insights into the system-level roles of non-coding regions in the human genome.

中文翻译：

---

静态非编码位点对染色质熵的差异影响


人类基因组的非编码区域对于功能调节很重要，但其机制仍然难以捉摸。我们使用机器学习来指导对中枢（即形成许多 3D 接触的非编码位点）进行 CRISPR 筛选，并显着提高了细胞生长必需的中枢的发现率。我们发现必需和非必需枢纽之间没有明显的遗传或表观遗传差异，但我们观察到线性基因组中的一些邻近枢纽具有明显的空间接触和对细胞生长的相反影响。表观遗传静止区域中的一对这样的基因对基因表达、染色质可及性和染色质组织表现出不同的影响。我们还发现，删除必需枢纽会改变遗传网络活动，并增加染色质可及性的熵，比删除非必需枢纽造成的熵更严重，这表明它们对于维持有序的染色质结构至关重要。我们的研究揭示了对人类基因组中非编码区域的系统级作用的新见解。