These days, new Google products and code seem to appear on a weekly basis. Take, for example, Google Gears which takes advantage of SQLite, mentioned on nodalpoint recently. They certainly don't hang about at the Googleplex in Mountain View, California. Wouldn't it be great if Google applied some of that engineering expertise and agility to science and bioinformatics? Just imagine: we could have Google Metabolic Maps, a virtual globe of the cell for scientists everywhere...
Scientists have been drawing metabolic maps for a very long time, but unfortunately when it comes to charting and understanding metabolic pathways, we're still at the "here be dragons" stage of bio-cartography. I'm obviously not the first person to dream of this, but imagine maps of metabolic pathways looked more like Google Earth or Google Maps, than the old fashioned style maps, many life scientists will be familiar with. Now imagine just a little more, that these maps weren't just available on conventional screens, but we're given the Minority Report treatment, courtesy of Mr Bill Gates and his wizzy surface magic at Microsoft. Wouldn't that be great? Metabolic maps on an interactive tabletop computer. Just like Tom Cruise in the movies, we'd be able to effortlessly swish around metabolism (or the metabolome / proteome / genome / [insert-your-favourite]ome). Imagine if it was all open-source too, no boundaries, no passports...
Now, you may say that I'm a dreamer, but I'm not the only one [1,2,3].
References
- Zhenjun Hu, Joe Mellor, Jie Wu, Minoru Kanehisa, Joshua M. Stuart and Charles DeLisi (2007) Towards zoomable multidimensional maps of the cell Nature biotechnology 25 (5), 547-54. DOI:10.1038/nbt1304
- Hiroaki Kitano, Akira Funahashi, Yukiko Matuoka and Kanae Oda (2005) Using process diagrams for the graphical representation of biological networks Nature biotechnology 23 (8), 961-6. DOI:10.1038/nbt1111
- John Lennon and Yoko Ono (1971) Imagine
This work is licensed under a
Creative Commons Attribution-Noncommercial-Share Alike 3.0 License.
Comments
AVIS Viewer
Hi,
We developed a Google gadget viewer that implements some of the features you guys are talking about.
http://actin.pharm.mssm.edu/AVIS2
Avi
virtual metabolic maps
Actually, we're much closer to this than you seem to realize.
My group's Pathway Tools software is able to automatically generate metabolic map diagrams for any organism
with a sequenced genome. These diagrams are customized to reflect the metabolism of
that organism, as opposed to the old-style metabolic charts that are mosaics across all
metabolism. Here's one in the form of a PDF for Mycobacterium tuberculosis:
http://nar.oxfordjournals.org/cgi/data/34/13/3771/DC1/2
and here's one for Caulobacter crescentus:
http://nar.oxfordjournals.org/cgi/data/34/13/3771/DC1/1
Both are mentioned in this article:
"The Pathway Tools Cellular Overview Diagram and Omics Viewer"
http://nar.oxfordjournals.org/cgi/content/full/34/13/3771
Download Pathway Tools, and you can create them yourself for any organism
you like. (http://biocyc.org/download.shtml)
Web versions of these diagrams that are unfortunately not yet scalable are available
for 260 organisms through the BioCyc query page: http://biocyc.org/server.html
Just select your organism of interest, and then click on the word Diagram in the fifth
bullet down this page, to see the diagram.
Give us 10% of Google's software development budget, and we'll be even
farther along...
BioCyc scalability
Hi Peter, I guess its you. Thanks for reminding me of BioCyc, I'm interested in what the issues are with scalability of the web versions, do you mean it takes "two to four minutes" to load yeast metabolism in BioCyc for example. Can't you just kidnap some google engineers? They only work just down the road from you :)
BioCyc scalability
Sorry if I wasn't clear about scalability. I was not talking about speed (usually the diagram of the
full metabolic network -- the cellular overview -- downloads faster than that), but about changing
the scale of the diagram -- zooming it. The Web version cannot be scaled; the version produced by
our downloadable software can be scaled.
It's probably just a matter of time before we get to that point. We have literally hundreds of
requests for enhancements from our user community to prioritize.
Very interesting
BioCyc sounds very interesting to me [especially you have mentioned the lisp api] and I would like to use. I visited the http://biocyc.org/download.shtml page and found that there is simply too much information needed to be put in to get the software. More-over what about hackers who are affiliated to some IT company and work on Bioinformatics stuff in their spare time... should they pay the fees to hack around the BioCyc?
______________________"The Answer Lies in Genome"______________________
http://fuzzylife.org/
Reply
Gosh, yes, ten fields to fill out! It could take a minute!
People working at IT companies who are interested in hacking on the software
in their spare time should contact us via ptools-support at ai dot sri dot com.
Lack of patience
Sorry about that, open-source-software movement and particularly the repo sf.net has spoiled me.
Thanks for the email id.
What is the Google magic?
To make the dream come truth, shall we pray for the Google magic? Yes... No ... Maybe ...?
Wait a minute, what exactly is the magic from Google?
AJAX? - No
Talented engineers? - No
Market opportunity - Maybe
Funding? - Maybe
What else? -- It is a combination of all the four above.
Now I understand why people talking "Click Your Heels Three Times and Bow to the Google Gods" and "takes a moment to imagine a shiny, happy world where the Google geniuses offer metabolic maps."
In other words, no NIH funded projects will be able to achieve this goal; only a deep-pocketed commercial entity is able to do it.
I agree that the metabolic map or a singling map of the cell is complex, but not impossible.
Simon
The beginning
AJAX? - No
Talented engineers? - Yes
Market opportunity - Hope so
Funding? - Hope so
Since then the algorithm has not changed much.
Complexity of Biological Data
Like ejain, we are also working on a metabolic pathways database and visualization tool, PathCase, at CWRU. Unlike Google's ajax-based solution, we embedded into our web interface a java applet that retrieves pathways from a web service and draws it on the client side.
I like the idea of adding a little bit "Google magic" to biology. In our case, unfortunately, metabolic maps are much more complicated than geographical maps. First of all, there is no common standard for drawing a metabolic pathway. For example we draw metabolites as yellow disks, whereas ExPASy uses free form labels for metabolites. Second, different organisms have alternative versions of the same pathway. Since we have more than 400 species (see KEGG's organism list), and the number is increasing everyday, we need to find an elegant way of displaying a pathway for different organisms. Third, individual pathways are usually separated spatially and temporally. When we merge pathways in a very large map, we may lose information. For example, let's say two pathways A->B->C and M->C->D are exclusive: A,B and M,D are found in different cell compartments, and C is found everywhere in the cell. Merging the two pathways generates an alternative pathway A->B->C->D which may not be observed in reality...
Many other problems can be enumerated as well. My point is that we currently have more problems with biological data than just visualizing it in a user friendly way.
Biology is more complicated than Geography
I agree, metabolic maps are much more complicated than geographical maps. Visualising them is a hard and largely un-solved problem, which is what makes it interesting. As for the lack of common standards for drawing metabolic pathways, perhaps the Systems Biology Graphical Notation (SBGN) will take us some of the way there. Thanks for the link to Pathcase, I took a look at the Glycolysis pathway in Pathcase, it would be nice if you could link to the map directly. I wonder, is there a limit to the number of nodes you can display? Tools like CellDesigner handle more than about 30 nodes quite badly...
Hi Duncan, Thank you for
Hi Duncan,
Thank you for your interest on SBGN, and also referring to our paper :-)
I have one question on your post above. Which version of CellDesigner did you use?
We have tuned the performance of CellDesigner on 3.1, so it want get awful performance degradation.
Although CellDesigner is not scalable enough, I think it might work for 30-50 nodes on current version
of CellDesigner (3.5.2)...
CellDesigner performance
Hi Akira (Funahashi) I guess its you? I was using CellDesigner 3.5.1, rendering 30-50 nodes sounds better...
Biological Network Visualization
SBGN is a nice start for standardization of pathway notations. I hope Dr. Kitano's group get enough funding to further grow SBGN as strong as its brother SBML.
Currently we are trying to provide stable links to individual entity pages (such as gene, enzyme, pathways) at PathCase. We managed to make the site indexable by search engines. In the next release we will hopefully get rid of dynamically generated identifiers at the URLs.
I am not sure about the maximum number of visible nodes in our applet. It mostly depends on memory size accessible by the applet. However, panning/editing the graph starts to slow down after inserting 300 nodes in a modern PC.
Re: Biology is more complicated than Geography
Biology more complicated than geography ? I'm not a geographer but I would say the opposite: geography has four dimensions (latitude, longitude, altitude, time) and many layers : topology, hydrography, roads, etc... Our poor genomic sequence has only one dimension, but there is still not a 'google earth-like' for genomics.
Geography vs. Biology?
What I meant by geographical maps was street maps that you can find printed versions at a Walmart. You sure can make any visualization/illustration complicated by increasing the detail level. In comparison with geographical maps, visualization of biological maps is just at its infancy. Geologists have been improving maps of Earth for hundreds of years. An accurate biological map has been available to biologists only for a few decades.
Many more than one dimension
I disagree. I wasn't just thinking of DNA, there's all the data (layers, if you like) for RNA, proteins, metabolites, enzyme kinetics etc that you might want to visualise. You already have the same four dimensions as geography, for example in environmental genomics and then you've got microscopic geography to think about as well (cellular location, compartments etc). To my mind, this makes visualising pathways very challenging, probably more so that simple geography.
Re: Google Metabolic Maps
Funny, I'm just experimenting with adding some Google magic to the "old fashioned style" pathway maps: http://eric.jain.name/pathways/map.html :-)
Not dissing Expasy
Hi Eric, I like your demo. Any plans to extend it? And why does Aldehyde Reductase get special treatment?
Re: Not dissing Expasy
Just waiting for a digitized version of the new, 4th edition of Roche's pathway "wall chart". The "Aldehyde Reductase" label serves as a test for linking items to the map...