Sixty-Four Free Chemistry Databases Part 31: Surfing Biological-Chemical Networks with STITCH

Chemistry databases become much more interesting when combined with other kinds of data. Today's stop our continuing tour free chemistry databases and Web services is another case in point. STITCH allows you to visualize and explore relationships between small molecules and biomolecules. From the about page:

STITCH is a resource to explore known and predicted interactions of chemicals and proteins. Chemicals are linked to other chemicals and proteins by evidence derived from experiments, databases and the literature.

STITCH contains interactions for over 68,000 chemicals and over 1.5 million proteins in 373 species.

STITCH treats the interactions between small molecules and biomolecules as a network that can be interactively explored. Each biological or chemical entity is a hub that's connected to other entities through some form of evidence.

For example, let's say we'd like to learn something about the enzyme Fatty Acid Amide Hydrolase (FAAH). We'd start by typing a term "fatty acid amide hydrolase" into the field marked "name" in the top left of the screen. Pressing "go" takes us to screen in which we can select from a number of species. Keeping the selection on Homo sapiens and pressing "continue" takes us to a screen in which the centerpiece is an interactive biological-chemical network.

We can see from this network that the FAAH enzyme (in red) is related to a number of other biomolecules and small molecules. For example, we might want to get a closer look at one of FAAH's endogenous substrates, anandamide (AEA). Clicking the shape labelled "AEA" activate the pop-up from which one option is to re-center the network around this small molecule. Selecting our species interest (Homo sapiens) takes two another network view with a anandamide in the center. We can now see that anandamide interacts with a number of other biomolecules. We can continue exploring this way until we found everything were looking for.

We might be interested in knowing why it is that anandamide is associated with the 5-HT7 receptor. We can answer this question by hovering over the line connecting the two entities a clicking. This takes us to a page summarizing the evidence for a functional link. In this case the evidence consists of several simultaneous mentions in a PubMed abstracts (which is rated low by the system). We can find out what these abstracts are by clicking on the "show" button.

I found some minor problems with STITCH. For example, it was difficult to get to the evidence view because detection of a mouse hover over an evidence line was quite sensitive. Performing a substructure search was not as intuitive as it could be because I was also asked to select a species in addition to providing a SMILES string. Finally, the structure for anandamide given by STITCH was way off:

There's much more to STITCH than what's been covered here but this brief introduction gives you a feel for what's possible. For anyone interested in small molecules and how they interact with biomolecules STITCH is one of the better resources I've come across.

Kudos

  • Interactive, engaging representation of complex data set.
  • Step-by-step, interactive mini tutorial.
  • Keep updated on new developments through STRING/STITCH blog.
  • Assertions can easily be corroborated through the primary literature.

Ideas for Improvement

  • Improve the usability of structure search by not requiring a species to be selected and by providing a chemical structure editor.
  • Implement a convenient process for correcting errors such as the error in the structure of anandamide.

Comments

  1. Michael Kuhn
    October 01, 2009 @ 7:26 AM

    Hi Rich,

    thanks for the kind review of STITCH. We're actually in the process of releasing the 2nd version, which is now in beta. Your concerns are totally justified. We've taken some steps in the right direction with the new version, though: you don't have to select a species when searching for a chemical anymore, and we at least show all of the different structural scaffolds that we have merged. I'm not quite sure where the totally wrong structure for anandamide came from, but even in the new version I discovered a bug and traced it back to an erroneous annotation at KEGG. I've alerted the KEGG guys, but before the corrections finds it way to us via PubChem, it'll take quite a while. :( We're a bit slow there because we're based on a frozen version of PubChem.

    In the long run, we want to provide better distinction between the different isomers we merged. But we'll still be susceptible to errors in PubChem that our heuristics don't catch.

    Anyway, details on the new features will soon appear on the blog and in the NAR 2010 db issue.

    best, Michael

Your thoughts?

No HTML. To create a link like this example, use: [example](http://example.com). Learn more