"Thinking traces" in large reasoning models: Cognitive cost or performative scaffolding?

大型推理模型中的“思维痕迹”:认知成本还是功能性支架?

阅读:1

Abstract

Long non-coding RNAs (lncRNAs) interact with chromatin and recruit epigenetic complexes to specific genomic loci, yet their relationship with super-enhancers (SEs), key regulatory elements frequently reprogrammed in cancer, remains unexplored. We developed an integrative pipeline that combines RNA–chromatin contact data (RNA-Chrom), histone modification–lncRNA expression correlation profiles (HiMoRNA peaks), and super-enhancer annotations (SEdb 3.0) to map lncRNA–SE regulatory axes. Applying this framework to SNHG1 in HCT116 colorectal cancer cells, we identified 21 SNHG1-reactive super-enhancers (Ψ-SEs) among 184 cancer-specific SEs, at which SNHG1 physical contacts co-occur with SNHG1-correlated histone modifications (HiMoRNA peaks), predominantly H3K4me1 (permutation p = 0.001, fold enrichment = 2.03). Comparison with 4145 lncRNAs demonstrated that epigenetic correlations alone do not distinguish SNHG1; instead, the addition of the contact layer is required to delineate the Ψ-SE set. Differential expression (DESeq2) and co-expression analyses in 471 TCGA-COAD tumor samples identified 12 Ψ-SE target genes (including CDC20, PDP1, and TOP1) consistently upregulated in both HCT116 cells and patient tumors and positively correlated with SNHG1, with the co-expression signal robust to tumor purity correction. The proposed Ψ/Ω classification provides a generalizable framework for prioritizing super-enhancers at which lncRNA–chromatin interactions may shape the local epigenetic environment across cancer types.

特别声明

1、本页面内容包含部分的内容是基于公开信息的合理引用;引用内容仅为补充信息,不代表本站立场。

2、若认为本页面引用内容涉及侵权,请及时与本站联系,我们将第一时间处理。

3、其他媒体/个人如需使用本页面原创内容,需注明“来源:[生知库]”并获得授权;使用引用内容的,需自行联系原作者获得许可。

4、投稿及合作请联系:info@biocloudy.com。