Introduction to the transomics2cytoscape package

Version Information

R version: 4.4.1

Bioconductor version: 3.19 (RCy3: 2.24.0)

Cytoscape: 3.10.3

Cy3D (Cytoscape app): 1.1.3

KEGGscape (Cytoscape app): 0.9.2

Introduction

Visualization of Trans-omic networks helps biological interpretation by illustrating pathways where the signals are transmitted (Gehlenborg et al., 2010).

To characterize signals that go across multiple omic layers, Yugi and colleagues have proposed a method for network visualization (Yugi et al., 2014) by stacking multiple 2D pathways in a 3D space.

The 3D network visualization was realized by VANTED (Rohn et al., 2012). However, the visualization relies on time-consuming manual operation. Here we propose transomics2cytoscape, an R package that automatically creates 3D network visualization in combination with Cytoscape (Shannon, 2003), Cy3D App, and Cytoscape Automation (Otasek et al., 2019).

Installation

This package requires Cytoscape to be installed and you need to run Cytoscape before running the following R code.

BiocManager::install("transomics2cytoscape")

Workflow

transomics2cytoscape has 2 main functions create3Dnetwork and createTransomicEdges Below is a diagram of the transomics2cytoscape workflow.

Figure 1. The transomics2cytoscape workflow
Figure 1. The transomics2cytoscape workflow

create3Dnetwork has 3 arguments.

The 1st one is a directory path where you put the network files to be layered in 3D space. The 2nd one is a file path of TSV for the Z-axis layout of the network files (called “Layer definition file”). The last one is a file path of XML used to style Cytoscape.

For example,

suid <- create3Dnetwork(networkDataDir, networkLayers, stylexml)

createTransomicEdges has 2 arguments.

The 1st one is the SUID of the network created by create3Dnetwork. The 2nd one is a file path of TSV for the transomic interactions (called “Transomic interaction file”).

For example,

suid <- createTransomicEdges(suid, layer1to2)

Input files

(Any number of) network files to be layered in 3D space

transomics2cytoscape can layer all the networks that Cytoscape can import. You need to put these files in the directory of the 1st argument of create3Dnetwork. You don’t need to put files for the KEGG pathway. For KEGG pathway, you can import the network just by writing the KEGG pathway ID in the “Layer definition file” described later.

Layer definition file

“Layer definition file” is a TSV file for the Z-axis layout of the network files.

A file that defines network layer index and the Z-height of the network in 3D space. The format is as follows.

kinase-layer    rno04910    1500    false
enzyme-layer    rno01100    900 true
metabolite-layer    rno01100    1   false

The 1st column is the network layer index. This information is added to the node table column LAYER_INDEX.

The 2nd column is the KEGG pathway ID or the network file name in the directory of the 1st argument of create3Dnetwork. You don’t need to prepare a network file for the KEGG pathway. You can import the KEGG pathway simply by writing the KEGG pathway ID.

The 3rd column is the Z-height of the network.

The fourth column specifies whether you want to connect the interaction to the “edge” of the network in that row. If this is true, transomics2cytoscape will create a node at the midpoint of all edges of the network. This is useful when you want to represent an interaction that activates and inactivates a reaction in the network. For example, KEGG global metabolic pathway network does not have an enzyme node, unlike a normal metabolic pathway. This column should be set to “true” in such cases.

A style file of Cytoscape

A Cytoscape style file. For more information about Cytoscape style file, see the Cytoscape user manual. Note that you can only use style properties that are supported by Cy3D.

Trans-omic interaction file

“Trans-omic interaction file” is a TSV file that defines the edges that connect the different network layers. The format is as follows.

Figure 2. Trans-omic interaction file
Figure 2. Trans-omic interaction file

The 1st ~ 3rd columns are the information about the node at the “source” of the transomic interaction.

The 4th ~ 6th columns are about the target node.

The 1st and 4th columns are the network layer index of the source/target node.

The 2nd and 5th columns are the column names for which you want to find the attribute value of the source/target node.

The 3rd and 6th columns are the attribute values of the source/target node should have.

The last column is the type of the transomic interaction. This information is added to the interaction column of the edge table.

Example

You can reproduce Figure5 of Yugi 2014 with the code below.

The execution of this code took 3 minutes to complete on a Windows 11 Desktop PC [12th Gen Intel Core i7-12700 2.10 GHz, 128 GB RAM, and Nvidia GeForce RTX 4070].

(Note: Do not operate Cytoscape until the code execution is completed.)

# suppressPackageStartupMessages(library(dplyr))
# suppressPackageStartupMessages(library(RCy3))
# suppressPackageStartupMessages(library(KEGGREST))
# Sys.setenv(LANGUAGE="en_US.UTF-8") 
library(transomics2cytoscape)
networkDataDir <- tempfile(); dir.create(networkDataDir)

networkLayers <- system.file("extdata/usecase1", "yugi2014.tsv",
                            package = "transomics2cytoscape")
stylexml <- system.file("extdata/usecase1", "yugi2014.xml",
                            package = "transomics2cytoscape")
suid <- create3Dnetwork(networkDataDir, networkLayers, stylexml)

layer1to2 <- system.file("extdata/usecase1", "k2e.tsv",
                            package = "transomics2cytoscape")
suid <- createTransomicEdges(suid, layer1to2)

layer3to2 <- system.file("extdata/usecase1", "allosteric_ecnumber.tsv",
                            package = "transomics2cytoscape")
ec2reaction(layer3to2, 6, "allosteric_ec2rea.tsv")
suid <- createTransomicEdges(suid, "allosteric_ec2rea.tsv")

Then, you should have a 3D view with layered networks and transomic interactions between them. (Note that you need to perform operations such as zooming out or adjusting the camera angle.)

Figure 3. Reproduction of Yugi’s transomics visualizaion (Yugi 2014)
Figure 3. Reproduction of Yugi’s transomics visualizaion (Yugi 2014)

Conversion from EC number to KEGG reaction ID

For those who have seen the enzyme reaction database such as BRENDA (Chang et al., 2021), it is not intuitive that the ID of the allosteric regulatory target (6th column of the allosteric_ecnumber.tsv) is the ID of the metabolic reaction rather than the EC number.

This is because KEGG uses the reaction ID instead of the EC number as the ID of the pathway object of the global metabolism map.

So transomics2cytoscape has a function ec2reaction that converts the EC number column of the Trans-omic interaction file into the KEGG reaction ID.

Figure 4. How the ec2reaction function works
Figure 4. How the ec2reaction function works
ecnum <- system.file("extdata", "allosteric_ecnumber.tsv",
    package = "transomics2cytoscape")
ec2reaction(ecnum, 6, "allosteric_ec2rea.tsv")

Example (with more network layers)

transomics2cytoscape can visualize more layers than in the previous example.

You can reproduce a visualiztion of Kokaji 2020 with the code below. (Unfortunately, this paper is not open access and I can’t give you the details.)

The execution of this code took 5 minutes to complete on a Windows 11 Desktop PC [12th Gen Intel Core i7-12700 2.10 GHz, 128 GB RAM, and Nvidia GeForce RTX 4070].

(Note: Do not operate Cytoscape until the code execution is completed.)

This visualization shows that transomics2cytoscape can also visualize network not only with the previous 3 layers but also with the 5 layers (and more than that).

networkDataDir <- tempfile(); dir.create(networkDataDir)
tfs <- system.file("extdata/usecase2", "TFs.sif",
                            package = "transomics2cytoscape")
file.copy(tfs, networkDataDir)
networkLayers <- system.file("extdata/usecase2", "kokaji2020.tsv",
                            package = "transomics2cytoscape")
stylexml <- system.file("extdata/usecase2", "Kokaji2020styles.xml",
                            package = "transomics2cytoscape")
suid <- create3Dnetwork(networkDataDir, networkLayers, stylexml)

layer1to2 <- system.file("extdata/usecase2", "s2t_name_updated.tsv",
                            package = "transomics2cytoscape")
suid <- createTransomicEdges(suid, layer1to2)

layer2to3 <- system.file("extdata/usecase2", "t2e_name_updated.tsv",
                            package = "transomics2cytoscape")
suid <- createTransomicEdges(suid, layer2to3)

layer3to4 <- system.file("extdata/usecase2", "e2r_updated.tsv",
                            package = "transomics2cytoscape")
ec2reaction(layer3to4, 6, "e2r_updated_ec2rea.tsv")
suid <- createTransomicEdges(suid, "e2r_updated_ec2rea.tsv")

layer5to4 <- system.file("extdata/usecase2", "m2r_updated.tsv",
                            package = "transomics2cytoscape")
ec2reaction(layer5to4, 6, "m2r_updated_ec2rea.tsv")
suid <- createTransomicEdges(suid, "m2r_updated_ec2rea.tsv")
Figure 5. Reproduction of Kokaji’s transomics visualizaion (Kokaji 2020)
Figure 5. Reproduction of Kokaji’s transomics visualizaion (Kokaji 2020)
sessionInfo()

References

Chang, A., Jeske, L., Ulbrich, S., Hofmann, J., Koblitz, J., Schomburg, I., et al. (2021) BRENDA, the ELIXIR core data resource in 2021: New developments and updates. Nucleic Acids Research, 49, D498–D508.
Gehlenborg, N., O’Donoghue, S.I., Baliga, N.S., Goesmann, A., Hibbs, M.A., Kitano, H., et al. (2010) Visualization of omics data for systems biology. Nature Methods, 7, S56–S68.
Otasek, D., Morris, J.H., Bouças, J., Pico, A.R. and Demchak, B. (2019) Cytoscape Automation: Empowering workflow-based network analysis. Genome Biology, 20, 185.
Rohn, H., Junker, A., Hartmann, A., Grafahrend-Belau, E., Treutler, H., Klapperstück, M., et al. (2012) VANTED v2: A framework for systems biology applications. BMC systems biology, 6, 139.
Shannon, P. (2003) Cytoscape: A software environment for integrated models of biomolecular interaction networks. Genome Research, 13, 2498–2504.
Yugi, K., Kubota, H., Toyoshima, Y., Noguchi, R., Kawata, K., Komori, Y., et al. (2014) Reconstruction of insulin signal flow from phosphoproteome and metabolome data. Cell Reports, 8, 1171–1183.