Thursday, March 10, 2016

Introducing Pathway Collages...

Figure 1
Pathway Tools has long been recognized for the quality of our automatically generated individual metabolic pathway diagrams, which are intuitive to biologists, can be shown at varying levels of detail, and can be customized in various ways, including with the overlay of omics data. When a more global view is called for, our cellular overview diagram depicts the entire metabolic network for an organism, with capabilities for selective highlighting and overlay of omics data. However, to understand some biochemical situations, viewing a single pathway is insufficient, whereas viewing the entire metabolic network results in information overload. Pathway Collages, new in Pathway Tools version 19.5, are an attempt to bridge this gap, allowing users to create high-quality, customized, user-manipulable diagrams containing collections of user-specified pathways.

Pathway Collages can be explored and edited via the Pathway Collage Viewer web browser application. This application, implemented using the Cytoscape.js open-source JavaScript graph visualization library, supports panning, zooming, and all the editing and customization operations described in this post and the documentation embedded within the Pathway Collage Viewer itself. Feel free to experiment yourself with the example pathway collage online at http://biocyc.org/cytoscape-js/ovsubset.html?graph=example1&showHelp=T, or create your own following the instructions below.

Figure 2
Three example Pathway Collage figures are illustrated here. Figure 1 depicts a Pathway Collage consisting of four E. coli pathways overlaid with gene expression data. This diagram has already been manually adjusted by repositioning the pathways relative to each other and tweaking node font sizes and shapes. Metabolites that are shared between pathways are indicated by drawing connecting lines between them. 

Figure 2 shows a collage consisting of two E. coli pathways overlaid with predicted reaction flux data. In this diagram, rather than drawing connecting lines, compounds that are shared between the two pathways are merged, showing glycolysis flowing seamlessly into fermentation.
Figure 3

Figure 3 depicts a collage containing a larger number of pathways at a lower zoom level, so metabolite, enzyme and gene names are automatically suppressed (the font size of the pathway labels has been increased so those labels remain visible). In addition to manually repositioning pathways, merging some common nodes, and changing the default colors, some metabolites of interest have been highlighted in purple.

Now that you've seen what you can do with a Pathway Collage, how can you create one for yourself? Pathway Collages can be created from either the BioCyc website (or other Pathway Tools-based website) or from desktop Pathway Tools. There are five basic steps.
  1. Specify the set of pathways to be included. The simplest and most reliable way to specify a set of pathways is to generate a SmartTable containing the desired pathways, and then export the SmartTable to a Pathway Collage. This works both for the desktop and web versions of Pathway Tools, and enables you to keep your list of pathways around in case you ever want to edit it or regenerate your collage. There are other ways to specify a set of pathways, such as by interactively clicking on them in the cellular overview diagram (desktop only), from an omics dataset (web only), or by creating a seed collage from a single pathway and then interactively adding more (web only). We may add additional options to specify pathways in the future. Consult the documentation for more details.
  2. Export to Pathway Collage Viewer. Pathway Tools will compute automatic layouts of the individual pathways within the collage, then position those diagrams next to one another horizontally, and send that initial layout of the collage to the Pathway Collage Viewer application in your web browser.
  3. Interactively refine and customize the collage. This can involve repositioning items, showing connections, adding, deleting or merging elements, editing labels, highlighting elements of interest, and/or customizing node and edge styles. By default, only the metabolites along the main backbone of a pathway are included in the diagrams, but side metabolites can be added interactively. Additional pathways involving a metabolite of interest can also be added interactively.
  4. Import omics data to be visualized on the collage (optional). Omics data can be added either before or after the collage is generated. The collage can display omics data associated with either genes, metabolites, or reactions. When multi-timepoint gene expression data is displayed, the display of enzyme names is suppressed.
  5. Save or export the collage. At any time, a pathway collage can be saved as a JSON-format graph file on your computer; that file can later be loaded back in to the collage viewer (not all browsers support this operation --- we recommend using Chrome or Firefox). A pathway collage can also be exported to a PNG-format image file for use in presentations or publications. The image will be generated with a resolution comparable to that of the display at the time the image is created (up to some maximum), therefore, the highest-quality images are obtained if the collage is displayed at a high zoom level when exporting.
For more information on Pathway Collages, see the Pathway Tools Website User Guide or the help documentation within the Pathway Collage Viewer itself.

12 comments:

  1. Very useful tool. Can we add expression profiles of isozymes from different tissues on single route? How about adding reversible enzymes?

    ReplyDelete
    Replies
    1. Thanks for your comment. The enzymes displayed in the pathways are the same ones you will see in the Cellular Overview Diagram for that organism -- they are not tissue-specific or organized by tissue, and if there are too many isozymes, I believe the display will cut them off after some number (perhaps we can tweak that number or eliminate it in future versions, if there is demand -- I believe currently the display will show up to 5 enzymes that have experimental evidence, but only up to 2 computationally predicted ones). Any enzyme that gets included in the collage can have expression data associated with it. Unfortunately, if an isozyme is omitted (because there are too many isozymes to display), its expression profile will also not be displayable.

      I'm not sure what you mean by adding reversible enzymes -- any enzyme associated with the reaction will be included in the display, subject to the numerical limits mentioned above. If a reaction is shown as reversible in our existing pathway and cellular overview diagrams, then it will be shown as reversible in the pathway collage. It is usually left up to the curators of the pathway to explicitly mark a reaction reversible in the diagram, since many reactions are technically reversible but drawing them all as such would obscure the direction of pathway flow. In the desktop software (not on the web version), you can set a preference to show all reversible reactions as such (instead of the default pathway flow direction) -- to be honest I'm not sure if that preference also carries over to the pathway collages, but if not that would be a relatively easy fix to make.

      We welcome suggestions and feature requests for this and our other tools, but can't guarantee we'll get around to all of them promptly!

      Delete
    2. More regarding the tissue profile question: if you have the same enzymes in different tissues, you could generate an expression dataset such that data from different tissues occupies different columns in the same datafile. Then you could import that dataset so that instead of having, say, 4 different boxes for a gene representing different timepoints in a time series, they would represent different tissues (it's all just numbers to Pathway Tools -- we don't care where those numbers come from). If the enzymes are not the same, there are other ways to finesse the data, such as by associating the expression values in each data column with the reaction in a given tissue instead of with the enzyme (the key in the first column of the input datafile would then be the EC# or reaction ID), but that would require more complex up-front manipulation of the input datafile (it's easy to generate a mapping between genes and reaction ids using SmartTables, but perhaps not so straightforward to do the substitution in your datafile).

      Delete
  2. What tool did you used to create these diagrams? Is it Creately online diagramming and collaboration software?

    ReplyDelete
    Replies
    1. Read the post. It is the Pathway Tools Pathway Collage tool.

      Delete
  3. Hi,

    You mentioned that you might alter the software so that all annotated isozymes can be viewed with omics data (rather than just 2 for computationally annotated enzymes). I would find this useful. Is there already a plan to do this?

    ReplyDelete
    Replies
    1. It's on my list, but I haven't looked into it yet. If it's easy to do I might be able to squeeze it in for our December release, otherwise it will have to wait.

      Delete
    2. Hi Suzanne, thanks, that would be really useful. Could I ask how Pathway Collage chooses the two isozymes to show? This becomes important when I want to display omics data.

      Delete
    3. Hi Suzanne, thanks, that would be really useful. Could I ask how the selection of which isozymes are shown is made?

      Delete
    4. This did turn out to be easy to do, so it will be in our December release. Currently, isozymes with experimental evidence are favored over those with only computational evidence, but other than that, the selection is arbitrary.

      Delete
  4. Thakyou very much:) Is there a date set for the release?

    ReplyDelete
    Replies
    1. I'm hesitant to give an exact date. We're currently looking at early December, but our dates have been known to slip depending on what comes up.

      Delete