Note: Currently this is just a straight copy of Robert's email, and does not reflect subsequent email discussions.

Queries over the MIR

Below are a collection (partial) of some canned SQL queries ovre the MIR. It is all based upon the model of provenance in the Information Model. I hope to be able to "show off" the personalisation aspects of myGrid. There are many more queries I could add to this list. What would be good in the myGrid demo is to have a set of canned and labelled queries that change the "orientation" of the data in the MIR view according to my focus of attention. definition of terms:

1. Project: I have a "Grave's disease project;
2. Experiment: I run many kinds of experiments within myGrave's disease project
3. Runs: each experiment may be run many times, for different data; also as I'm notified to changes affecting the data in my project.

Queries:

• Show me my projects and related entities
• Show me the experiments and related entities for project x.
• show me the history of runs for experiment x
• What experiments has person x run.
• Who owns/manages prpject x
• Who created experiment x
• What were the results of running experiment x at time y.
• Show me the stats of usage of resource x.
• What services to I commonly use.

Each of these may be filtered: Show me projects managed by x, at time y, run by z, etc. these all have semantic variants:

• Show me projects (across or within projects) to do with human SNPs.
• What data do I have concerning GPCRs.
• What experiments do I do with task x.

We have a model of provenence (embedded in the information model). With each of these queries, I'm essentially changing the focus of the model. So, when I ask for an experiment, I expect to see a collection of "experiments" with all the "stuff" to do with an experiment: Creator, user, inputs, outputs, intermediates, components, services, times, dates, performance,... Each query "personalises" my view over the MIR to the way I wish to organise and view those data. You may wish to think of it as something like the "view" options in the Window Explorer. I can arrange by name, type, date, time, etc. Two issues (at least) remain:

1. populating the MIR with enough stuff (projects, experiments, users, ....) to make it look good.
2. Attaching semantic descriptions to these data.

Yours Robert.

Related

RobertsISMBDemoScript -- MarkGreenwood - 30 Apr 2003 See also MirQueries -- ChrisGreenhalgh - 12 May 2003