Systems Biology Layout & Rendering Service (SyBLaRS) is a web service to lay out graphs in SBGNML, SBML, GraphML and JSON formats and/or produce corresponding images (in JPG, PNG or SVG formats) of the layouts with an option to highlight results from various graph queries in the backend.
Main capabilities of SyBLaRS include:
- creating an image of the provided map, which already has layout information, with an option to highlight a specific query result
- laying out the provided map in specified layout style (among many available ones) and returning the map with layout information in JSON format, and
- both laying out the provided map in specified layout style and creating an image of it again with an option to highlight a specific query result (and returning both the map with layout information and the image).
Backed by these capabilities, SyBLaRS can be used:
- to produce images from a dataset with large amount of graphs (e.g. BioModels) in an automated way (see test branch)
- as the layout service of a network visualization tool, and
- to generate image of an SBGN/SBML model for a publication (e.g. journal/conference article).
SyBLaRS is distributed under the MIT License. Here is a sample server deployment along with a simple client-side demo:
Please cite the following when you use this service:
H. Balci, U. Dogrusoz, Y.Z. Ozgul and P. Atayev, "SyBLaRS: A web service for laying out, rendering and mining biological maps in SBGN, SBML and more", PLOS Computational Biology, 18(11), pp. 1-12, 2022.
In order to deploy and run a local instance of the service, please follow the steps below:
git clone https://github.com/iVis-at-Bilkent/syblars.git
cd syblars
npm install // this may take a while
The default port is 3000, you can change it by setting 'PORT' environment variable.
npm run start
Note #1: We recommend the use of Node.js version 14.x and npm version 6.x. We used Node.js v14.18.0 and npm v6.14.8 during development.
Note #2: This service uses Puppeteer to generate the output. Please refer to the Puppeteer documentation to ensure that your machine is configured properly to run Chrome headlessly.
Alternatively, you can use Dockerfile provided in the root directory. Please note that Dockerfile currently works in Linux and Windows environments and does not work in macOS because of Puppeteer related issues. To run the Dockerfile (below commands may require sudo in Linux environment):
First, cd
into the folder where Dockerfile is located.
Then, build a Docker image with name syblars (this may take a while).
docker build -t syblars .
Lastly, run the image from port 3000. If you want to use another port, please change the first port number in below command.
docker run -p 3000:3000 syblars
SyBLaRS supports the following input formats for graphs: SBGNML, SBML, GraphML, and JSON.
The notations used for these formats:
SBGNML Stylesheet | SBML Stylesheet | GraphML & JSON |
---|---|---|
The supported graph layout algorithms are: fCoSE, CoLa, CiSE, Dagre, Klay, Avsdf and are among Cytoscape.js layout extensions as listed here.
Each layout style has a varying number of options for customization of the layout. The demo provided only exposes some of the popular options; please refer to the corresponding GitHub repository for an exhaustive list of such options. Among these options some that we think will be commonly used are listed below:
padding
: Padding around the map drawingidealEdgeLength
: Ideal length of an edge (layout will try to get every edge this long within constraints)randomize
: Use random node positions at the beginning of layout (false
means an incremental layout)packComponents
: Whether or not to pack disconnected components after separate layouttile
: Enable tiling of disconnected nodes together for a compact representationnodeDimensionsIncludeLabels
: Whether or not to include labels in node dimensions during layout
The supported graph queries are:
- Degree Centrality: The normalized degree centrality of the nodes in the graph
- Closeness Centrality: The normalized closeness centrality of the nodes in the graph
- Betweenness Centrality: The normalized betweenness centrality of the nodes in the graph
- Page Rank: The rank of the nodes in the graph
- Shortest Path: The shortest path from a single source to a single target in the graph
- k-Neighborhood: The neighbors of the specified source nodes within a certain distance k
- Common Stream: The set of common nodes that are in the upstream/downstream/bothstream of all specified source nodes with a path length limit k
- Paths Between: The subgraph that consists of the paths of length at most k between any two nodes of the specified source nodes
- Paths From To: All shortest paths between specified source nodes and target nodes with a maximum length limit k and a further distance d
For more details about these queries, please refer to centrality section of Cytoscape.js and cytoscape.js-graph-algos GitHub repository.
Sending request to the local deployment via curl:
curl -X POST -H "Content-Type: text/plain" --data "request_body" http://localhost:3000/file_format
and via Fetch API
let settings = {
method: 'POST',
headers: {
Accept: 'application/json',
'Content-Type': 'text/plain'
},
body: request_body
};
let result = await fetch("http://localhost:3000/file_format", settings)
.then(response => response.json())
.then(res => {
return res;
})
.catch(e => {
return e;
});
where file_format
is one of the sbgnml
, sbml
, graphml
and json
.
request_body
needs to be formed in the following way:
If the file content is in JSON format:
JSON.stringfy([JSON.parse(file_content), options])
If the file content is in other formats:
file_content + JSON.stringfy(options)
where options
is an object consisting of layoutOptions
, imageOptions
and queryOptions
. Example:
options = {
layoutOptions: {
name: 'fcose',
nodeDimensionsIncludeLabels: true,
...
},
imageOptions: {
format: 'png', // output format
background: 'transparent', // background color
width: 1280, // desired width
height: 720, // desired height
color: '#9ecae1' // node color
},
queryOptions: {
query: 'kNeighborhood', // query type
sourceNodes: [nodeID1, nodeID2], // source nodes
limit: 2, // path length limit
direction: 'BOTHSTREAM', // direction of the search
sourceColor: '#00ff00', // highlight color for source nodes
pathColor: '#ffff00', // highlight color for result elements
highlightWidth: 10, // underlay padding used to highlight elements
cropToResult: false, // whether to crop the image to the query result
}
}
Note: While sending the requests via curl, any "
in the request_body
should be replaced with \"
and all newline characters should be removed.
For supported layout options, please check the documentation of the associated layout algorithm. Image options support three output formats: png
, jpg
and svg
. background
attribute should be a hex color code or transparent
. color
attribute should be a hex color code for SBML, GraphML and JSON formats, and one of the following predefined color schemes for the SBGNML format: bluescale
, greyscale
, red_blue
, green_brown
, purple_brown
, purple_green
, grey_red
, and black_white
.
The query options change depending on the query type:
Degree Centrality:
queryOptions: {
query: 'degreeCentrality', // query type
direction: 'DIRECTED', // direction of the search
highlight: true, // whether to highlight nodes based on the resulting values
highlightColor: '#00ff00' // highlight color (valid only if highlight option is true)
}
Closeness Centrality:
queryOptions: {
query: 'closenessCentrality', // query type
direction: 'DIRECTED', // direction of the search
highlight: true, // whether to highlight nodes based on the resulting values
highlightColor: '#00ff00' // highlight color (valid only if highlight option is true)
}
Betweenness Centrality:
queryOptions: {
query: 'betweennessCentrality', // query type
direction: 'DIRECTED', // direction of the search
highlight: true, // whether to highlight nodes based on the resulting values
highlightColor: '#00ff00' // highlight color (valid only if highlight option is true)
}
Page Rank:
queryOptions: {
query: 'pageRank', // query type
highlight: true, // whether to highlight nodes based on the resulting values
highlightColor: '#00ff00' // highlight color (valid only if highlight option is true)
}
Shortest Path:
queryOptions: {
query: 'shortestPath', // query type
sourceNodes: [nodeID1], // source node - only one node id
targetNodes: [nodeID2], // target node - only one node id
direction: 'DIRECTED', // direction of the search
sourceColor: '#00ff00', // highlight color for source node
targetColor: '#ff0000', // highlight color for target node
pathColor: '#ffff00', // highlight color for resulting elements
highlightWidth: 10, // underlay padding used to highlight elements
cropToResult: false // whether to crop the image to the query result
}
k-Neighborhood:
queryOptions: {
query: 'kNeighborhood', // query type
sourceNodes: [nodeID1, nodeID2, ...], // source nodes
limit: 1, // path length limit
direction: 'BOTHSTREAM', // direction of the search
sourceColor: '#00ff00', // highlight color for source nodes
pathColor: '#ffff00', // highlight color for resulting elements
highlightWidth: 10, // underlay padding used to highlight elements
cropToResult: false // whether to crop the image to the query result
}
Common Stream:
queryOptions: {
query: 'commonStream', // query type
sourceNodes: [nodeID1, nodeID2, ...], // source nodes
limit: 1, // path length limit
direction: 'BOTHSTREAM', // direction of the search
sourceColor: '#00ff00', // highlight color for source nodes
targetColor: '#ff0000', // highlight color for common nodes in stream
pathColor: '#ffff00', // highlight color for resulting elements
highlightWidth: 10, // underlay padding used to highlight elements
cropToResult: false // whether to crop the image to the query result
}
Paths Between:
queryOptions: {
query: 'pathsBetween', // query type
sourceNodes: [nodeID1, nodeID2, ...], // source nodes
limit: 1, // path length limit
direction: 'DIRECTED', // direction of the search
sourceColor: '#00ff00', // highlight color for source nodes
pathColor: '#ffff00', // highlight color for resulting elements
highlightWidth: 10, // underlay padding used to highlight elements
cropToResult: false // whether to crop the image to the query result
}
Paths From To:
queryOptions: {
query: 'pathsFromTo', // query type
sourceNodes: [nodeID1, nodeID2, ...], // source nodes
targetNodes: [nodeID1, nodeID2, ...], // target nodes
limit: 1, // path length limit
furtherDistance: 1, // parameter for "relaxing" the path length limit
direction: 'DIRECTED', // direction of the search
sourceColor: '#00ff00', // highlight color for source nodes
targetColor: '#ff0000', // highlight color for target nodes
pathColor: '#ffff00', // highlight color for resulting elements
highlightWidth: 10, // underlay padding used to highlight elements
cropToResult: false // whether to crop the image to the query result
}
where direction
is one of UPSTREAM
, DOWNSTREAM
or BOTHSTREAM
for kNeighborhood
and commonStream
queries and one of DIRECTED
or UNDIRECTED
for degreeCentrality
, closenessCentrality
, betweennessCentrality
, shortestPath
, pathsBetween
and pathsFromTo
queries.
In case you do not need layout applied, you should either not provide layoutOptions
or provide preset
the layout style. If you do not provide imageOptions
, default ones will be used. To disable image output (in case you only need the output JSON file back with layout information), you should set image
option to false
in your request URL, which is true
by default:
http://localhost:3000/file_format?image=false
After the request is sent, the server will lay out the given graph and return the layout information in JSON format that will contain node IDs along with their corresponding positions (x, y) and dimensions (width, height), and image information (in base64uri
encoding for png
and jpg
formats and in xml
for the svg
format). If you want the edges to be returned as well in layout information, you should set edges option to the request URL, which is false by default:
http://localhost:3000/file_format?edges=true
If an error occurs, the response of the server will consist of an error message.
For instance, a sample SBGNML file can be laid out with fCoSE layout and a corresponding PNG image can be generated by making a query to the sample deployment of SyBLaRS web service via curl in the following way:
curl -X POST -H "Content-Type: text/plain" --data "<?xml version='1.0' encoding='UTF-8'?><sbgn xmlns='http://sbgn.org/libsbgn/0.3'> <map version='http://identifiers.org/combine.specifications/sbgn.pd.level-1.version-1.3' id='map1'> <glyph class='simple chemical' id='glyph1'> <label text='DHA-P'/> <bbox x='30' y='20' w='60' h='60'/> </glyph> <glyph class='simple chemical' id='glyph2'> <label text='GA-3P' /> <bbox x='30' y='220' w='60' h='60'/> </glyph> <glyph class='macromolecule' id='glyph3'> <label text='Triose-P
Isomerase' /> <bbox x='150' y='120' w='120' h='60'/> </glyph> <glyph class='process' orientation='vertical' id='pn1'> <bbox x='50' y='140' w='20' h='20'/> <port x='60' y='130' id='pn1.1'/> <port x='60' y='170' id='pn1.2'/> </glyph> <arc class='production' source='pn1.1' target='glyph1' id='a1'> <start x='60' y='130' /> <end x='60' y='80' /> </arc> <arc class='production' source='pn1.2' target='glyph2' id='a2'> <start x='60' y='170' /> <end x='60' y='220' /> </arc> <arc class='catalysis' source='glyph3' target='pn1' id='a3'> <start x='150' y='150' /> <end x='70' y='150' /> </arc> </map></sbgn>{\"layoutOptions\":{\"name\":\"fcose\",\"randomize\":true,\"padding\":30},\"imageOptions\":{\"format\":\"png\",\"background\":\"transparent\", \"width\": 1280, \"height\": 1280, \"color\":\"bluescale\"}}" http://syblars.cs.bilkent.edu.tr/sbgnml?edges=true
and the corresponding response with image and layout information will be as follows:
{"image":"data:image/png;base64,iVBORw0KGgoAAAANSUhE...""layout":{"pn1":{"position":{"x":167.44176266064383,"y":152.28835944241527},"data":{"width":25,"height":25}},...}}}}
The same query can be done via Fetch API in the following way:
let settings = {
method: 'POST',
headers: {
Accept: 'application/json',
'Content-Type': 'text/plain'
},
body: <?xml version='1.0' encoding='UTF-8'?>
<sbgn xmlns='http://sbgn.org/libsbgn/0.3'>
<map version='http://identifiers.org/combine.specifications/sbgn.pd.level-1.version-1.3' id='map1'>
<glyph class='simple chemical' id='glyph1'>
<label text='DHA-P'/>
<bbox x='30' y='20' w='60' h='60'/>
</glyph>
<glyph class='simple chemical' id='glyph2'>
<label text='GA-3P' />
<bbox x='30' y='220' w='60' h='60'/>
</glyph>
<glyph class='macromolecule' id='glyph3'>
<label text='Triose-P
Isomerase' /> <!-- contains line break -->
<bbox x='150' y='120' w='120' h='60'/>
</glyph>
<glyph class='process' orientation='vertical' id='pn1'>
<bbox x='50' y='140' w='20' h='20'/>
<port x='60' y='130' id='pn1.1'/>
<port x='60' y='170' id='pn1.2'/>
</glyph>
<arc class='production' source='pn1.1' target='glyph1' id='a1'>
<start x='60' y='130' />
<end x='60' y='80' />
</arc>
<arc class='production' source='pn1.2' target='glyph2' id='a2'>
<start x='60' y='170' />
<end x='60' y='220' />
</arc>
<arc class='catalysis' source='glyph3' target='pn1' id='a3'>
<start x='150' y='150' />
<end x='70' y='150' />
</arc>
</map>
</sbgn>{"layoutOptions":{"name":"fcose","randomize":true,"padding":30,"nodeDimensionsIncludeLabels":true,"uniformNodeDimensions":false,"packComponents":true,"nodeRepulsion":4500,"idealEdgeLength":50,"edgeElasticity":0.45,"nestingFactor":0.1,"numIter":2500,"tile":true,"tilingPaddingVertical":10,"tilingPaddingHorizontal":10,"gravity":0.25,"gravityRange":3.8,"gravityCompound":1,"gravityRangeCompound":1.5,"initialEnergyOnIncremental":0.3},"imageOptions":{"format":"png","background":"transparent","width":1280,"height":720,"color":"bluescale"}} // file_content + JSON.stringfy(options)
};
let result = await fetch("http://syblars.cs.bilkent.edu.tr/sbgnml?edges=true", settings)
.then(response => response.json())
.then(res => {
return res;
})
.catch(e => {
return e;
});
let imageInfo = result["image"]; // data:image/png;base64,iVBORw0KGgoAAAANSUhE... (in `base64uri` for `png` and `jpg` and in `xml` for `svg`)
let layoutInfo = result["layout"]; // {"_00ac7e0a-288f-42e0-b252-9bcb59027572":{"position":{"x":97.42656942899578,"y":-128.0473968715957},"data":{"width":25,"height":25,"parent":"_23d436e2-6633-42f2-98d0-00dbb962ac3d"}},...}}}} (in JSON format)
See also here for sample Python codes that call SyBLaRS to lay out and render SBGNML files by Augustin Luna.
-
SyBLaRS regards any node position information. This is especially useful if you want to create an image of the map, which is already laid out. This is also useful in cases where you do have a partially decent layout but you would like to apply an incremental layout respecting the current positions in your map. This information should be provided via
bbox
of each glyph in SBGNML files, via the layout extension in SBML files, viax
andy
data attributes andposition
attribute of each node in GraphML and JSON files, respectively (see the sample files). -
SyBLaRS also takes any node dimensions into account for layout in SBML, GraphML and JSON file formats. This information should be provided via the layout extension in SBML files, and via
width
andheight
data attributes of each node in GraphML and JSON files (see the sample files). Any dimension data is ignored for the SBGNML file format, however, as the third-party stylesheet used for rendering SBGNML maps has fixed/pre-defined dimensions for each node type. -
Compound/nested graph structures are supported in all file formats. We recommend that you use the fCoSE layout style for best results on graphs with compound structures.
-
Performing a layout emphasizing the clustering/grouping of nodes is supported only with GraphML and JSON formats. The cluster each node belongs to should be defined via the
clusterID
data attribute of each node (see the corresponding sample files). CiSE layout style should be used for layout to explicitly show the clustering available in the graph. -
The
x
andy
coordinates of a node in the resulting layout information indicate the 'center' coordinates of the node. -
jpg
output format does not support transparent background; thus, we return a white background whentransparent
option is chosen. -
The layout options presented in the client demo are not exhaustive and may not include all options of the corresponding layout style. Please refer to the webpage of each layout extension for the detailed list of available options.
SyBLaRS uses the Express framework for handling HTTP requests. Actual operations are performed using Cytoscape.js and its extensions (see the package.json
file for a complete listing). Among these extensions, Cytosnap is particularly needed for creating a headless Chrome instance, on which graph creation, rendering, layout and image creation of the input graphs are performed.
Icons in the client demo are made by Freepik and Flaticon licensed with Creative Commons BY 3.0.
Third-party libraries used in web service: sbgnml-to-cytoscape, cytoscape-sbgn-stylesheet, cytosnap, libsbmljs, express, cors, jQuery, jsdom, nodemon, jest, super-test
Third-party libraries used in demo client: Semantic UI, underscore.js, backbone.js, FileSaver.js
Hasan Balci, Ugur Dogrusoz, Yusuf Ziya Özgül and Perman Atayev of i-Vis at Bilkent University