As part of the molecular goniometers project, Raghu will be doing a bit of protein engineering, and Shawn and I adopted one specific task that Raghu will be doing (fusing GFP and BurrH) to reimagine in Realtalk.  We did a couple weeks of paper mockups [2], and a first pass at a Realtalk prototype [3].

Protein Kit first prototype:  [Protein Kit first prototype.mov]

Our immediate goal is to record another video of ourselves engineering a protein in Realtalk as we "wish we could".  What's shown this second video will not be immediately realizable in Realtalk as it is, but should be "just beyond" what's currently possible.  The video can then serve as a "spec".  We (all of Team Realtalk) can identify which of the interactions in the video require improved capabilities from Realtalk, and come up with a plan to implement those capabilities.  Eventually, we will record another video which looks like the first one, except we are doing the task for real instead of as a prototype.

(On the other hand, as I was messing around with this stuff yesterday, the friendly IT guy dropped by the space with some visitors in tow, and I was able to give an impromptu and very well-received demo, which felt really good.  So there's also something to be said for building up more of a demo portfolio with Realtalk-as-is, before switching to an improving-Realtalk phase.  We'll probably end up doing both in parallel.)

((Also -- after I wrote this, I received the invitation to the Foresight Institute meeting in July, and I haven't yet talked with Shawn about this, but if it's a real thing, it could be a really good target for a demo showcase, which might affect priorities.))

It has been surprising and gratifying to start to get a glimpse of an integrated Realtalk-based environment for the entire scientific process.  Our earlier proposals had three headings ("augmented wet lab", "design and analysis tools", "learning and communicating"), but now those mostly seem like a reflection of how the scientific process is currently partitioned.  As Shawn and I have been going through this exercise, a more seamless vision has been emerging, where the representations that you are working with in the design stage are tied directly to the experiments that you'll be performing in the wet lab, the materials in your freezer and available from vendors, the data you'll be collecting and how you'll be viewing it, the paper that you'll publish...  This integration is (to me) still just a glimpse and hard to explain in words, but it feels like a big deal and also very achievable.

I drew a kind of "map" of the process, as seen through the goniometer project [1].  What we are doing right now is prototyping one corner of the map ("protein design"), but everything connects to everything else, and we can imagine further prototypes staking out other corners, until we've eventually connected enough dots to be able to demonstrate a beginning-to-end Realtalk-based workflow.

[1]

[2]

[3]