rm-plugin-part2

This video demonstrates the construction of the RM portion of our classification workflow.

The finished workflows collects a GEO dataset, uploads it RA, reformats the data, splits the dataset in half, trains a classifier on one half, tests the classifier model on the other half, then reports the classifier performance in the users web-browser

1. use rm-taverna plugin to browse repository and choose where to load data from for 

2. add 2 "replace" operators, which re-write the label attribute values (from Mutant-1 to Mutant, and Control-2 to Control etc.) this shows the operator config panel

3. Show how to connect up services (operators) in taverna

4. add 2 "set role" operators to set the "id" and "label" roles on the data

5. add "split data", shows the slight hack for handling dynamic outputs, 2 comma delimited output locations leads to 2 output ports

6. add an "SVM" and "performance" operator

7. add a beanshell to pop up the performance data in the users web browser

8. show how taverna can collect lots of different outputs from all parts of the workflow (e.g. the CSV we uploaded, the list of assay IDs, the URL sent the web-browser etc.