Welcome to EpiMogrify !!!

EpiMogrify is a computational method to prioritize signalling molecules for cell maintenance and cell conversion.

EpiMogrify models the cell's epigenetic state using H3K4me3 histone modifications (ChIP-seq peak breadth) and integrates protein-protein interaction network to predict the regulators of cell identity. We systematically identify signalling molecules that facilitate the development of serum-free chemically defined cell maintenance and differentiation media. Finally, this resource provides a catalogue of signalling molecules for cell maintenance and cell conversion, for more than 100 cell types made available by ENCODE consortia.

I. Cell Maintenance

EpiMogrify priortizes signalling molecules like receptors and ligands for cell maintenance in vitro . The receptors are produced by the cell type of interest whereas, the ligands can be produced by the cell type itself or support cell types. The top predicted ligands from the ranked receptor-ligand pairs could be added to the culture condition for in vitro cell maintenance.

II. Cell Conversion

EpiMogrify prioritizes signalling molecules such as receptor-ligand pairs for directed differentiation from pluripotent cells into terminally differentiated cell types or for enhanced cell conversion protocols. The predicted ligands can be added to the cell conversion protocol.

For further information and to cite: Uma S. Kamaraj, Joseph Chen, Khairunnisa Katwadi, John F. Ouyang, Yu Bo Yang Sun, Yu Ming Lim, Xiaodong Liu, Lusy Handoko, Jose M. Polo, Enrico Petretto and Owen J.L. Rackham EpiMogrify models H3K4me3 data to identify signaling molecules that improve cell fate control and maintenance. Cell Systems (2020). doi: 10.1016/j.cels.2020.09.004

Please use the below form to select the cell type of interest that you wish to culture and click 'submit'. The predicted list of ligands with the corresponding target receptors are displayed. The prioritised ligands can be added or coated to culture the cells in vitro . In addition, you can also choose the support cell types (user-defined) that might produce ligands to maintain the cell of interest.

Please use the below form to select the source cell type and target cell type from the drop-down list that you wish to view and then click 'submit'. The predicted list of ligands with the corresponding target receptors are displayed. The top predicted ligands could be added to the cell conversion protocol for cell conversion from source to target cell type.

Please use the below form to enter HUGO gene symbol of interest to view the H3K4me3, H3K27me3 histone modifications or gene expression values across all cell types used in this study.

H3K4me3 ChIP-seq peak breadth or height values associated with the gene across all cell types.

H3K27me3 ChIP-seq peak breadth or height values associated with the gene across all cell types.

RNA-seq gene expression in TPM (Tags per million) values across all cell types.

1. How does EpiMogrify predict ligands for cell maintenance?

In order to culture cells in vitro , it is important to mimic the environment in which they are found in vivo . For instance, even if ex vivo cells are expressing the required receptors, the microenvironment to stimulate these receptors can be lost in vitro . Therefore, in order to predict the best microenvironment for cell maintenance in vitro , we use EpiMogrify to rank receptor-ligand pairs and subsequently, these predicted ligands are added to the culture media.
EpiMogrify uses a data-driven approach to model the cell identity genes using associated broad H3K4me3 ChIP-seq peaks. We found that genes associated with H3K4me3 peak breadth greater than 87% of the peak breadth distribution (ranging from 3,611 to 15,035 base pairs depending on the cell type) efficiently marks the cell identity genes. EpiMogrify models the H3K4me3 peak breadth and gene-regulatory network (STRING database) to identify cell identity genes and signalling molecules for cell maintenance.
Step 1:
For each gene, the Differential Broad Score (DBS) is computed by comparing the broad H3K4me3 breadth values between the target cell type of interest and background cell types. DBS is a composite score calculated as the difference in associated breadth value ( Δ Peak breadth) and the significance of this difference (P-value). The protein-coding genes are ranked by DBS.
Step 2:
To prioritize genes with regulatory influence, for each gene the Regulatory Differential Broad Score (RegDBS) is computed as the weighted sum of the gene's DBS and connected gene's DBS in the STRING protein-protein interactions (PPI) network. The cell type-specific RegDBS is used to predict (i) cell identity genes and (ii) signalling molecules for cell maintenance.
Step 3:
Next, EpiMogrify predicts signalling molecules like receptors and ligands for cell maintenance in vitro . The receptors are produced by the cell type of interest whereas the ligands can be produced by the cell type itself or support cell types. The receptor and ligand pairs are ranked based on the receptor's RegDBS and corresponding ligand's DBS values. The top predicted ligands from the ranked receptor-ligand pairs are added to the culture condition for in vitro cell maintenance.

2. How to use EpiMogrify website to get the ligands for cell maintenance culture conditions?

Please go to Cell Maintenance page and select the cell type of interest from the drop-down menu and press 'Submit'. If you would like to add support cell types based on prior knowledge that secretes ligands for the maintenance of the cell type of interest, then choose the support cell types from the given multiple-choice list. The displayed table shows the predicted ligands that can be used in maintenance culture conditions and the cell identity receptors it targets. The last column provides the cell types from which the ligand is produced.

3. How does EpiMogrify predict ligands for cell conversion?

For cell conversion, EpiMogrify models the H3K4me3 profile of the source and target cell types to identify the cell identity of the source and target cell types. Then, EpiMogrify combines this change in cell identity and integrates the regulatory network information to calculate a Reg Δ DBS. EpiMogrify was applied to prioritize receptor-ligand pairs for all pairwise conversions in the ~100 ENCODE/Epigenome Roadmap datasets by ranking receptors by Reg Δ DBS, ligands by Δ DBS and combining ranks to produce and order set of receptor-ligand pairs.
Step 1:
For cell conversion from source to target cell type, the change in cell identity is computed as the difference in DBS value between the source and target cell types.
Step 2:
Then a cell conversion Reg Δ DBS is calculated as the sum of the gene's Δ DBS and STRING protein-protein interaction network score. The genes are ranked by cell conversion Reg Δ DBS to predict signalling molecules for cell conversion. We then applied EpiMogrify to prioritize receptor-ligand pairs for all pairwise conversions in the ~100 ENCODE/Epigenome Roadmap datasets by ranking receptors by Reg Δ DBS, ligands by Δ DBS and combining ranks to produce and order set of receptor-ligand pairs. The top predicted ligands could be added to the cell conversion protocol.

4. How to use EpiMogrify website to find the ligands for cell conversion culture conditions?

Please go to Cell Conversion page and select the source cell type and target cell type of interest for cell conversion from the drop-down menu and press 'Submit'. The displayed table shows the predicted ligands that can be used in cell conversion conditions and the receptors it targets to facilitate the change in cell state.

5. How are EpiMogrify's predictions experimentally validated in cell maintenance and differentiation?

We demonstrate the power of this approach by comparing EpiMogrify predicted conditions with more typical chemically undefined conditions (such as Matrigel or Geltrex) in two ways. Firstly, we show that by using EpiMogrify-predicted factors for maintenance conditions, we were able to better potentiate the maintenance of astrocytes and cardiomyocytes in vitro . Secondly, we report a significant increase in the efficiency of both the differentiation of astrocytes and cardiomyocytes using EpiMogrify-predicted factors for conversion conditions. For more details, please refer to the paper.

6. How are receptors and ligands defined in this approach?

In this approach, the receptor-ligand interaction pairs consisting of 708 ligands and 691 receptors were obtained from a study (Ramilowski et al., 2015) and these pairs were identified based on the gene expression across different cell types. Further, the ligands that are known to be secreted based on UniProt database are predicted.

7. How to cite this approach?

Uma S. Kamaraj, Joseph Chen, Khairunnisa Katwadi, John F. Ouyang, Yu Bo Yang Sun, Yu Ming Lim, Xiaodong Liu, Lusy Handoko, Jose M. Polo, Enrico Petretto and Owen J.L. Rackham EpiMogrify models H3K4me3 data to identify signaling molecules that improve cell fate control and maintenance. Cell Systems (2020). doi: 10.1016/j.cels.2020.09.004

For further queries on the method and website: