Monday, 23 June 2008

Atlas Web Services Alpha


A new build is up - 4866. No major visible changes but several important reworkings under the hood:
  • Improved caching - hopefully you should notice faster responses
  • Better experimental factor ontology (structured vocabulary of experimental variables) expansion - now expands down through all available levels instead of just one as before, and is on by default. Example: search for tumor.
  • Auto-suggest drop-down works a bit more intuitively, esp. on conditions. You'll see what I mean, just start typing a query.
The big deal with this release is, however, the very limited, initial SOAP Web Services API to the Atlas. See for further detail on this. Among other things it's a step towards batch querying of data.

As always, your feedback is welcome! More interesting and wonderful things are in the works.

P.S. It would be easy for us to offer expansion by lots of other ontologies. Would that be useful to users, too?

Wednesday, 4 June 2008


A brief update on the ArrayExpress Atlas project. Build 4720 is up! It turns out the mailing list subscription form on the front page wasn't working. If you tried to subscribe and didn't hear anything back from us, please subscribe again.

ArrayExpress Experimental Factor Ontology

An experimental feature has been added. Now that our curation team has released an updated version of the ArrayExpress Experimental Factor Ontology (read the news item or browse the EFO ontology in EBI's OLS) we can use this to help your queries across experimental factor values (conditions) work better. For example, in EFO the "cancer" term has three children, "sarcoma", "chordoma" and "carcinoma". If you just query for cancer, you will get a certain number of hits, including "breast cancer" and "gastric cancer" and so forth.

Now, try to run the same query, turning on the EFO expansion checkbox. You will see that among the conditions you've found now there are such hits as "clear cell sarcoma of the kidney" and "bladder carcinoma", etc. The reason these conditions now are coming up is that we expanded your original "cancer" query with 1 level below it in EFO. Try it out.

Of course this just a first attempt at using the EFO ontology in this project. There are many issues that still need to be addressed, such as - how far into the ontology should we dig to expand our queries? 1 level? 2 or more? Also, what is the best way to present the results of an expanded query?

In any case, while we are working on improving this, we thought we'd let you try out this first approach already. Do let us know what you think!

All the best,

Misha Kapushesky
ArrayExpress Atlas Coordinator