The website begins with the 'Gene Pair Selection' section. Start by choosing a gene of interest. Following this, you'll be presented with gene pairs that have significant co-regulation with your selected gene. While you have the option to adjust the FDR cutoff to define the network's significance level, we suggest beginning with an FDR of \(10^{-6}\). By default, the edge list is organized in descending order of significance, from the most significant to the least.

Please select the gene of interest:



Gene Pair Edgelist [Please select a row below before moving on to 'Scatter Plot' or 'Network Plot']:

In the 'Gene Pair Edgelist' table, the first 2 columns indicate the gene symbols.

  1. The 3rd column 'P-adj Value' indicates how significantly the gene pair co-regulated.
  2. The 4th column 'Non-linear Rate' indicates whether two genes are non-linearly correlated.
  3. The 5th column 'Gene Pair Index' indicates a unique index for each gene pair.
Download Table (.csv)

*Non-linear Rate: Consider a gene pair, \( x \) and \( y \). If \( x \) and \( y \) always change in a concordant manner or always change in a discordant manner, then \( x \) is linearly correlated with \( y \), leading to a non-linear rate of \( 0 \). However, if \( x \) changes concordantly with \( y \) half of the time and discordantly the other half, the non-linear rate is \( 1 \).

A gene pair \( x \) and \( y \) are concordant if and only if when \( x \) goes up, then \( y \) goes up, or when \( x \) goes down, then \( y \) goes down. A gene pair \( x \) and \( y \) are discordant if and only if when \( x \) goes up, then \( y \) goes down, or when \( x \) goes down, then \( y \) goes up. In other words, if \( x \) and \( y \) move in the same direction, we call this pair concordant. If \( x \) and \( y \) move in opposite directions, we call this pair discordant.


On this page, two scatter plots are presented, showcasing the RNA expression level (\( \log(X + 1) \) transformed) of your selected gene pair. The first is a static plot, depicting samples from experiments where this gene pair exhibits significant co-regulation. In contrast, the second is an interactive plot. Each frame of this interactive version displays samples from an individual experiment where the gene pair is significantly co-regulated.

Scatter Plot of Selected Gene Pair:


'Circles' are samples from Positively Co-regulated Contrasts.

'Triangles' are samples from Negatively Co-regulated Contrasts.

Note: Samples belong to the same contrast are indicated by the same color. Due to the large number of co-regulated contrasts, samples from different contrasts may share similar colors.


Hovering over the samples will provide detailed expression data, with 'Gene1' represented on the x-axis and 'Gene2' on the y-axis. Click the 'play' button to automatically view the co-regulation of this gene pair across various experiments. You also have the option to manually navigate by dragging the progress bar.

Notably, for non-linear gene pairs, experiments showing concordant regulation typically appear before those displaying discordant regulation.


On this page, a network is displayed with your selected gene as the central hub. Genes most significantly co-regulated with this hub gene are connected as neighboring nodes. You can opt for the ‘Only Show Major Links’ view (set as default) to exclusively reveal connections between the hub gene and its neighboring genes. Alternatively, the 'Show Entire Network' option displays connections among the neighboring genes as well, though only those surpassing a set significance level will be presented. Use the left panel to adjust the number of top neighboring genes displayed; by default, the top 30 are shown.

Within the network:

  • Green Nodes represent Online Mendelian Inheritance in Man (OMIM) genes, signifying that the gene is associated with a known disease phenotype. Detailed information on these genes can be accessed in the subsequent ‘OMIM Table’. While this table lists all OMIM genes that are significantly co-regulated with the hub gene, only the most significant ones are visualized in the network.
  • Blue Nodes denote genes with clinically relevant variants from ClinVar. These genes potentially have links to diseases but are not fully validated. In the ‘ClinVarCount Table’ that follows, the column labeled ‘ClinVarVC’ indicates the number of disease-related variants associated with each gene. A gene with a higher count of variant records in ClinVar is more likely to be linked to a disease."
  • Khaki Solid Edges denote a linear correlation between two genes.
  • Violet Dashed Edges denote a non-linear correlation between two genes.

Note: It's essential to ensure that the ClinVar database is indeed used for indicating potentially disease-linked but not validated genes, as described.

Please choose one of the below:



OMIM Table: OMIM genes co-regulated with the hub gene

ClinVarCount Table: ClinVar Genes co-regulated with the hub gene



On this page, you can initiate an analysis to identify which gene ontology (GO) terms are enriched among the top co-regulated genes in the network. The analysis allows selection of up to 500 of the most significantly co-regulated genes. These selected genes will appear in the 'Gene Sets' box below.

Additionally, you have the option to manually input specific gene symbols or ENSEMBL IDs for analysis. Please note that running the GO analysis might take between 1 to 2 minutes.

Gene Ontology Enrichment Test on Top Co-regulated Genes

The analysis process may take 1-2 minutes

Data description

The co-regulation network is generated based on Recount3 human data: https://rna.recount.bio/

We collected \(4,473\) human experiments with a sample size between \(6\) and \(40\) from the Recount3 portal. We successfully extracted group label information in \(3,293\) experiments. For the remaining experiments, we used the DASC tool to identify hidden labels in \(458\) experiments. According to the label information, we found \(19,681\) total contrasts of which \(18,997\) were from manual labels and \(684\) were from DASC labels. These studies contained \(56,498\) total samples. We used DESeq2 to identify genes with fold changes greater than \(1.5\) and FDRs smaller than \(0.01\) as DEGs. Finally, we collected \(12,082\) contrasts with DEG numbers between \(2\) and \(3,000\), and \(27,342\) genes which have been identified as DEGs in at least \(5\%\) of contrasts, as our input data.

R package for creating co-regulation network: https://github.com/Jiasheng-Wang/CoRegNet