1. Install python packages:

   pip install -r requirements.txt

   Note that if you have problems installing cython or jnius-indra, which are dependencies of indra, you can ignore them as this demo doesn't need that functionality.

That's it!

First, you'll need an instance of Sbgnviz-Collaborative-Editor running at localhost:3000. The server address is hardcoded at the moment, but it should be trivial to modify the code to change that if necessary. The room ID is provided as a command-line argument, though:

python agent.py <room_id>

The client will report on the users currently in the room and echo any chat messages it sees. There will also be some debug prints from indra which we've neglected to remove and can be ignored.

Once the agent, "INDRA", has connected to the Editor, submit text to be assembled by writing it in the chat with the prefix "indra:". For example:

indra: EGFR binds EGF. EGFR bound to EGF phosphorylates itself at Y1068.

The agent will acknowledge that it has accepted text for processing, and then make another announcement when the processing is complete.

Note that the parser used behind the scenes is slow. Even five or six simple sentences could take upwards of one minute to process. Also keep in mind that the canvas is cleared before each new model is loaded, i.e. it does not add nodes incrementally to an existing map.