Documents and tutorial

This page provides the complete tutorial for CohesinDB.

The main contents of CohesinDB include three types of cohesin objects (i.e., cohesin binding sites, cohesin related chromatin loops and cohesin related cis-regulatory modules) and the processed data involved (e.g., bigwig files for ChIP-seq.):

• Cohesin objects: Browse, Analyze, Statistic, Download
• NGS data involved: Visualize, Data, Statistic
• Others: Home, Documents

Additional information 1: You can jump to the section you are interested in by selecting the panel on the right -->>

Additional information 2: Documentation on the code and maintenance of CohesinDB can be found here. Notify datasets to CohesinDB can be found here

Additional information 3: Please use this page to contact us for any bugs or problems (or contact wangjk[AT]m.u-tokyo.ac.jp).

The cohesin-mediated loops are identified from 119 cohesin ChIA-PET data, 42 cohesin Hi-ChIP data and 385 Hi-C data. Overall, CohesinDB includes ~1 million (957,867) chromatin loops. Similar to 'browse cohesin sites', to browse cohesin-mediated loops: (1) you enter this cohesin loop page by click the dropdown list. (2) Then you can enter filters to define the criterias for cohesin loops. (3) After clicking the submit button, the right panel (4) shows the overview of search results. The table (5) gives the main information of the selected loops. You can check the information and annotations in this table, or enter the details page via click on the CDB ID (6) You can download the table results as a csv or Excel format.

* Too few filters will make this webpage heavy.

By click on the CDB ID (CDBL*), you can enter the details pages for each cohesin loops. You can share the URL of detailed pages to others such as collaborators The detailed page provides annotations including:

• Basic information

A unique identify; Genomic position of the loop; Loop length.

• Data source

The assay type (Hi-C, Hi-ChIP or ChIA-PET); Involved cohesin subunit; Cell types of the involved samples; Related studies.

• Related cohesin and genes

Cohesin sites and genes that overlap with the selected loop.

• Visualization

The plot of selected loop and related cohesin sites and related genes. You can drag to move the coordinates. This section is based on the embedding of WashU Epigenome Browser

This page is for cohesin-regulated genes. CohesinDB considered two types of evidence to obtain cohesin-gene pairs.

• Differential expressed genes identified from Transcriptomic data (e.g. RNA-seq, cohesin-knockdown compare with control). Only DEGs identified in ≥ 2 studies are used. Cohesin sites less than 2mb from DEGs are considered as regulatory cohesin.
• Chromatin loops that connected cohesin sites to gene promoters.

Overall, CohesinDB contains 2,229,500 double-evidenced cohesin-gene pairs.

The user interface is the same with cohesin sites and loops. (1) You enter this cohesin-mediated genes page by click the dropdown list. (2) Then you can enter filters to define the criteria for genes. (3) After clicking the submit button, the right panel (4) shows the overview of search results. The table (5) gives the main information of the selected genes. You can check the information and annotations in this table, or enter the details page via click on the CDB ID (6) You can download the table results as a csv or Excel format.

By click on the CDB ID (CDBG*), you can enter the details pages for each gene:

• Basic information

CohesinDB ID; Gene symbol; Ensembl ID; Gene locus; Strand; Gene type.

• Double-evidenced regulatory sites

Whether cohesin-regulated gene; Evidence types; regulatory cohesin sites; Co-expression with cohesin genes.

• Tissue specificity

Gene expression levels (TPM) across tissues. The data are from GTEx

Tissue-specificity (Gini coefficient) of genes

Cell-specificity of regulatory cohesin.

• Cohesin loop evidence

Whether connect cohesin via cohesin loops; Involved cell types; Connected loops; Loop type (Direct means cohesin sites are located on gene body); Loop study; Loop subunit

• Cohesin DEGs evidence

Whether cohesin DEGs; Involved cell types; Studies; Cohesin subunits.

• Visualization

The plot of selected gene and related cohesin sites and loops. This section is based on the embedding of WashU Epigenome Browser

This analyze page is designed for biologist who don't need to know computational knowledge. The analysis includes three part:

Analyze1: Predict regulatory elements

Input a list of genes, to obtain the regulatory sites and co-bind transcriptional factors.

Analyze2: Predict target genes

Input a list of regions, to obtain the target genes and gene pathway.

Analyze3: Loop discovery

Input a list of genomic regions, to obtain the related loops and the connected regions (the other end of the loops).

** Due to the extensive role of cohesin, these analyses are useful not only for cohesin research, but also for any study on gene regulation / chomatin interaction / 3D genome. For example, cohesin is involved in almost all cis-regulatory elements [Riku, Nat Gen, 2015]; Cohesin plays a major role in chromatin organization [Rao, Cell, 2017]

This analysis takes a gene set as input, then output the regulatory sites with co-bound TFs. This analysis can be performed by selecting the specified cell type, or by disregarding the cell type (All celltype). The input should be a list of gene symbol in HGNC format. Because one gene can be regulated by many genomic loci where many TFs located, the plot only shows partial results: (1) cohesin sites are merged; (2) Cohesin sites that involved in many (top 10) cohesin-gene pairs are shown; (3) The most frequently involved (Top 10) TFs are shown. The full results are provided with a .csv file which contains 3 columns:

Gene	Regulatory-Cohesin	Cobind-TFs
PAK4	chr19-38150586-38152046	ZFX,SP140,ZBTB48,ESR1...

This analysis takes a list of genomic regions as input, then output the candidate genes that are regulated by the input regions. The input should be comma-separated genomic regions under hg38 reference.This analysis can be performed by selecting the specified cell type, or by disregarding the cell type (All celltype). Similarly, the plot only shows partial results: (1) Genes that are involved in many (top 10) cohesin-gene pairs are shown; (2) Gene pathway with the most (top 10) significant p-value are shown. The full results are provided by two downloadable files:

Sites-Genes-Pathway

Input sites	Target genes	Significant_enriched_pathway
chr12-116515022-116516937	MED13L	Thyroid hormone signaling pathway

The full list of KEGG enrichment

Gene_set	Term	Overlap	P-value	...
KEGG_2021_Human	Biosynthesis of unsaturated fatty acids	1/27	0.0094131205361766	...

This analyze is to find chromatin loop for a list of genomic loci. The output are revelant chromatin loops and the connected regions by the loops. Rather than treating chromatin as a linear model (i.e., nearby regions (±5kb) of interest loci), this analysis enables researchers to study genomic loci of interest in a higher-dimensional way. For example, the connected regions may far (>2 mb) away from the input regions, which are not considered in the linear model.

The input file should be comma-separated .bed format. Only the first 3 columns are used. This analysis can be performed by selecting the specified cell type, or by disregarding the cell type (All celltype). The plot on the right panel shows the most (to 10) relevant study (ChIA-PET, Hi-ChIP or Hi-C). The downloadable files include discovered loops (columns 1-6: two anchors of loops, column 7: studies.) and the connected regions (columns 1-3: input regions; columns 4-6: connected regions).

This page provides the visualization for ChIP-seq, ChIA-PET, Hi-ChIP and Hi-C included in CohesinDB.

Basically, this page is based on the API (embedding) provided by WashU Epigenome Browser. The interface is the same with WashU browser.

Try this page by clicking the example.

To use this page. (1) Select the sample of interest. Assay -> Cell -> Sample. (2) Click 'Add' button. The chosen sample will be added to the current sample list. (3) Check current samples or clear all selected data. (4)(5) Check the gene annotation and genome coordinates. You can move the genes/coordinates by 'Clicking and Dragging'. Or you can use the tool bar on the top. (6) Check the visualization of selected samples. You can re-arrange the panels by dragging them.

Of note, this "Visualize" page is designed to visualize the NGS dataset used in CohesinDB. CohesinDB does not act as a server for the visualization of external data. Instead, users can find the URLs for each dataset in CohesinDB, by right-clicking on each track (see picture below). Then, this URL for CohesinDB data, together with the user's own data, can be used in other specialized web servers such as WashU Epigenome Browser or UCSC genome browser.

By click on the result in the table, you can access the detailed pages for each data or each cell type.

The detailed information for each data include:

Basic information: ID, cohesin-related type, assay type, involved cohesin subunit.

Sample information: Cell type, Tissue type, Biosample, Related disease, Treatment.

Assay-specific information: Antibody, Restriction enzyme, Comparison group.

Downloadable file: Link to the original study, download link.

The descriptions of downloadable files are:

Assay	Contents	Format	Format descriptions
ChIP-seq	ChIP-seq reads	.bigwig	An a popular binary format to represent numerical ChIP-seq reads values over genomic coordinates. Definition and usage can be found in UCSC
ChIP-seq	ChIP-seq Peaks	.narrowPeak	Called peaks of signal enrichement based on normalized data. It is a BED 6+4 format and details can be found here
ChIA-PET	Chromatin loops	.longrange	A bed format-like file type which contains 4 columns from left to right: chromosome, start position, and end position, interaction target. It can be used in WashU Epigenome Browser
ChIA-PET	Chromatin loops	.bedpe	A file format based on the BED format to concisely describe disjointed genomic features such as chromatin loops. Developed by the bedtools team;
Hi-C, Hi-ChIP	Hi-C contact matrix	.hic	An indexed binary format designed to permit fast random access to contact matrix heatmaps. The format was designed by the Juicer
Hi-C, Hi-ChIP	Chromatin loops	.bedpe	A file format based on the BED format to concisely describe disjointed genomic features such as chromatin loops. Developed by the bedtools team;
Hi-C	Topologically associated domain	.tad	It is the similar with bedpe format, which mainly include the genomics loci of the left and right boundaries of TADs.
RNA-seq etc.	Differential expressed genes.	tsv	Results of edgeR. It mainly include the information of logFC and FDR of a gene.
RNA-seq etc.	Gene expression (TPM).	tsv	The results of STAR. It mainly include the TPM and FPKM of a gene.

This page summarized the key information for CohesinDB.

• Number of samples for each cohesin subunit. We used log2 to achieve a better visualization. The majority of them are Rad21, SA1, SMC3. Hi-C samples are assigned to 'Other' because Hi-C provides indirect information for cohesin. Samples with CdLS disease are assigned to NIPBL subunit.
• Number of samples for cell types. Overall, the top three cell types are MCF-7, HCT-116 and HeLa cells.
• Information of all cohesin sites. For example, almost all (>96%) cohesin sites also overlap with cohesin ChIA-PET loops, suggesting the reliability of CohesinDB.
• Information of all cohesin-regulated genes. Based on ~2 million cohesin-gene pairs, there are 15203 genes that are regulated by cohesin, with double-evidence.
• Studies included in CohesinDB.