Abstract:Silencers are repressive cis-regulatory elements that play crucial roles in transcriptional regulation. Experimental methods for identifying silencers are always costly and time-consuming. Computational methods, which relies on genomic sequence features, have been introduced as alternative approaches. However, silencers do not have significant epigenomic signature. Therefore, we explore a new way to computationally identify silencers, by incorporating chromatin structural information. We propose the SilenceREIN method, which focuses on finding silencers on anchors of chromatin loops. By using graph neural networks, we extracted chromatin structural information from a regulatory element interaction network. SilenceREIN integrated the chromatin structural information with linear genomic signatures to find silencers. The predictive performance of SilenceREIN is comparable or better than other states-of-the-art methods. We performed a genome-wide scanning to systematically find silencers in human genome. Results suggest that silencers are widespread on anchors of chromatin loops. In addition, enrichment analysis of transcription factor binding motif support our prediction results. As far as we can tell, this is the first attempt to incorporate chromatin structural information in finding silencers. All datasets and source codes of SilenceREIN have been deposited in a GitHub repository (https://github.com/JianHPan/SilenceREIN).

What problem does this paper attempt to address?

### What problem does this paper attempt to solve? This paper aims to solve the problem of identifying silencers in the genome. Specifically, the authors propose a new computational method - SilenceREIN - to find silencers at the anchors of chromatin loops. Traditional experimental methods for identifying silencers are both expensive and time - consuming, and existing computational methods mainly rely on linear genome or epigenome features, but these methods are not very effective for silencers because silencers lack prominent epigenome features. To solve this problem, the SilenceREIN method combines chromatin structure information and linear genome features. By using Graph Neural Networks (GNN), SilenceREIN extracts chromatin structure information in the regulatory element interaction network and integrates it with linear genome features to more accurately predict the location of silencers. This is the first attempt to incorporate chromatin structure information into silencer identification. ### Main research content 1. **Background introduction**: - Silencers are important cis - regulatory elements and are involved in transcriptional regulation. - Experimental methods for identifying silencers are costly and time - consuming. - Existing computational methods mainly rely on linear genome or epigenome features, but with limited effectiveness. 2. **Method innovation**: - Propose the SilenceREIN method, which combines chromatin structure information and linear genome features. - Use graph neural networks to extract the structural features of the regulatory element interaction network. - Integrate the extracted structural features with linear genome features to build a deep - learning model. 3. **Data processing**: - Collect regulatory element data such as silencers, enhancers, and promoters in the K562 cell line. - Integrate ChIA - PET data to construct a high - resolution regulatory element interaction network. - Extract histone modification signals and transcription factor binding signals as features. 4. **Model evaluation**: - Use 5 - fold cross - validation to evaluate the model performance. - Compare with the existing state - of - the - art methods (such as DeepSilencer and gkmSVM). - The results show that SilenceREIN performs better or equally well on multiple performance metrics. 5. **Result discussion**: - SilenceREIN can effectively identify silencers located at the anchors of chromatin loops. - Through APA analysis, the enrichment relationship between the silencers predicted by SilenceREIN and promoters is verified. - It is found that silencers show significant enrichment in CTCF and POLR2A ChIA - PET data, but no such enrichment is observed in H3K27ac HiChIP data, implying its negative correlation with active gene expression. ### Conclusion SilenceREIN is an innovative computational method. By combining chromatin structure information and linear genome features, it can more accurately identify silencers in the genome. This method not only improves the prediction accuracy but also provides a new perspective for understanding the function of silencers.

SilenceREIN: seeking silencers on anchors of chromatin loops by deep graph neural networks

Whole-genome Epigenetic Function Annotation Through Sgrna Libraries Synthesized by Controlled Template-Dependent Elongation

SilencerDB: a Comprehensive Database of Silencers

DeepICSH: a complex deep learning framework for identifying cell-specific silencers and their strength from the human genome

Chromatin Potential Identified by Shared Single Cell Profiling of RNA and Chromatin

A Chromatin Activity Based Chemoproteomic Approach Reveals a Transcriptional Repressome for Gene-Specific Silencing

DeepRegFinder: deep learning-based regulatory elements finder

Predicting active enhancers with DNA methylation and histone modification

A novel interpretable deep learning-based computational framework designed synthetic enhancers with broad cross-species activity

ScReNI: single-cell regulatory network inference through integrating scRNA-seq and scATAC-seq data

IReNA: Integrated Regulatory Network Analysis of Single-Cell Transcriptomes and Chromatin Accessibility Profiles

EPInformer: A Scalable Deep Learning Framework for Gene Expression Prediction by Integrating Promoter-enhancer Sequences with Multimodal Epigenomic Data

A comparison of experimental assays and analytical methods for genome-wide identification of active enhancers

DeepRCI: Predicting RNA-chromatin Interactions Via Deep Learning with Multi-Omics Data

Integration of Count Difference and Curve Similarity in Negative Regulatory Element Detection

scENCORE: leveraging single-cell epigenetic data to predict chromatin conformation using graph embedding

Expression of the Neural REST/NRSF-SIN3 Transcriptional Corepressor Complex as a Target for Small-Molecule Inhibitors

OCRFinder: a Noise-Tolerance Machine Learning Method for Accurately Estimating Open Chromatin Regions

Identification, Design, and Application of Noncoding Cis-Regulatory Elements

Genome-wide identification of stable RNA-chromatin interactions

Chromatin interaction neural network (ChINN): a machine learning-based method for predicting chromatin interactions from DNA sequences