1) Our user has just finished some Affymetrix microarray experiments to find genes which are differentially expressed in lymphocytes (a type of blood cell) from a set of patients with Graves disease compared to some healthy controls.

2) The user enters the data into the standard Affymetrix software and performs a query that will find all the genes that are significantly upregulated or downregulated in the dataset. The result of the query is a set of ‘genes’ identified by a unique identifier (assigned by Affymetrix), called an AffyId^?. For the sake of simplicity we can assume that each AffyId^? maps to a single gene on the human genome.

3) Our user now decides that she would like to learn more about the genes that are downregulated in the diseased patient. In particular she would like to view up to date annotation on the genes, obtain functional information and browse and display SNPs associated with the various genes in the dataset in order to design a Genotype experiment for genes of interest. She decides to use MyGrid, via the labBook system. She saves the list of AffyIds^? that identify the 85 downregulated genes as an ascii text file to her local hard disk.

4) She fires up the labBook application (A java application installed on the client?) running on her PC running Windows XP.

5) She is greeted by a splash screen welcoming her to the MyGrid labBook as the application is loaded and presented with a framework for a graphical interface, including menu structure etc. and is immediately prompted to logon to MyGrid. (see AlansLabBookStoryBoard (ALBSB))

6) On a successful logon the application maximises itself across the screen and is clearly split into frames (Similar to Netbeans layout – see also http://twiki.mygrid.info/twiki/bin/view/Mygrid/NetBeansLabBook). Within one of the frames the contents of her information repository are displayed. The information repository is displayed in a number of tabbed panes with the content categorised by type (e.g. Workflows, Projects etc.). See point 3 of ALBSB.

7) Our user has already created a project called ‘Graves Disease Gene Expression’ so she clicks the tab on the information repository browser to bring up the pane containing her current projects and selects ‘Graves Disease Gene Expression’

8) Details about the current project are displayed in a frame within the LabBook application showing the free text annotation that she entered when she created the project, together with existing data items associated with the project. Also listed are the previous ‘in silico’ experiments that she has done in the project.

9) Clicking on a experiment brings up details of the experiment including provenance data and links to the associated results sets in some kind of experiment browser window

10) The newly obtained list of AffyIds^? needs to be uploaded into this project. See AlansLabBookStoryBoard section 5 & 6. She is prompted to load a file from the disk. Ideally the labBook should automatically determine that the file contains a collection of AffyIds^? and associates the attribute ‘Collection’ with the data item. ‘Collections’ are automatically displayed in a Swing JTable when clicked on.

11) The users sees that a new data icon has appeared in the Project browser frame. She clicks on it to view its contents and the contents are displayed within a pop-up, scrollable JTable.

12) The user then decides that should like to perform a new ‘in silico’ experiment to analyse this new dataset and to examine the SNPs associated with the genes represented by the AffyIds^?. She intends to use three existing workflows that have been kindly provided for her by the good MyGrid people. (She may want to search for other workflows that are able to handle this type of data).

13) The first of these workflows provides annotation for the genes represented by the AffyIds^? and looks up the GO terms for the protein encoded by each gene. She creates a new experiment definition (myExpt2) that is associated with her project (Graves Disease Gene Expression) ‘with a title, free text description, plus the automatically trapped basic provenance metadata’. See ALBSB section 7. Possibly from a menu or from a menu brought up by right clicking in the Project Browser window or through the use of a wizard.

14) The experiment definition ‘attributes’ are displayed in some kind of dialogue box/form that permits:

• The input data source be chosen (possibly by dragging the icon from the project contents window)

[In this case the input dataset is a ‘collection’ of AffyIds^? – the workflow need to be run iteratively for each item in the collection. A biologist is not interested in running it individually on each item :-). A what level does this occur- the labBook? Portal? Enactment engine? Something has to recognise that the dataset is a collection]

• Allows the appropriate workflow be chosen (again by dragging from the MIR browser or by selecting from a list of appropriate workflows selected by matching ‘type’)
• Allows some free text annotation to be associated with the experiment
• Also possibly allows user to select an enactment engine. See ALBSB sect 9.

15) Should our user be happy to use the workflows as provided for this scenario ? Ideally she should be free to edit the workflow in order to:

• Click on a workflow, open it up in a workflow editor and select alternative services from a list.
• Change the default parameters associated with the services.
• Specify at which points on the workflow data, is to be recorded in the MIR
• Save the modified workflow to the MIR.

16) She would now like to perform the experiment by running the workflow. She right clicks on the new experiment definition that has now appeared in the project browser and chooses run from the menu. [Note: Do we assume here that at experiment definitions are distinct from instances of experiments (experiments that have been run)? Both could appear in the project browser but have different icons]

A run dialogue box appears asking for confirmation and details about whether and how she would like notification of the experiment completion (e.g. email or info dialogue box within the labBook)

17) After notification of completion of the experiment a new icon representing an instance of an experiment appears in the project browser window. Our user wishes to view the experimental results in collective, predefined view. She clicks on the experiment instance and a table appears that displays columns that include the AffyId^?, the appropriate EMBL accession number for each AffyId^?, The Swissprot ID if available for each AffyId^? , A list of Medline Ids, and the GO term if available for the protein in the cross referenced SwissProt^? ID. The table is sorted by GO term in order to cluster AffyIds^? that map to similar GO terms together.

[At this point we assume some magic – how does the labBook know which of the data items from the results of the workflow to display? The table display will probably need to be hard coded for the demo. In the future should there be a wizard for the display of the data contained within instances of experiment?]

18) The results of this experiment now constitute a pool of ‘genes’ that she would like to work with individually. She browses the table showing the results of the first instance of myExpt2 and decides that she would like to run another workflow (the second one provided by MyGrid) on the second gene listed in the table in order to view the SNPs that lie within that gene, and to design some primers for a genotyping experiment.

[At this point we have another challenge. How can we select an item from a displayed collection and turn it into a stand alone data instance suitable for input to another workflow?]

She points to the second row and right clicks on an appropriate cell in the table (either an AffyId^? or the EMBL id?) and selects 'Create new Data Item' from a menu. A new data item now appears in the project explorer pane with the correct type that will now act a data input for workflow 2.

To Be Continued..

-- AnilWipat - 10 Apr 2003