ART proteins are active on their cognate arrays
Okay, the funding may be a barrier to many; this experiment used $14K at catalog prices. Thanks to the efficiency of our autonomous laboratory infrastructure, we can have thin margins after passing through consumable costs, but those consumable costs are still frustrating. DNA synthesis is a major component, and still too expensive (and too slow!). But even with the barrier higher than we'd like, it's much lower than before. Put another way, the number of individuals or companies that could have executed this experiment in our Cloud Lab is orders of magnitude greater than those who have both the appropriate laboratory equipment and the appropriate lab skills.
I wasn't one to criticize Anthropic for the lack of experimental data; many of my first author academic publications are purely computational. Once in a while I was able to entice an experimentalist to follow-up, but that was the rarity. Didn't help that my first paper had over 400 observations across similar numbers of bacteria; there's no easy way to validate at that scale.
Now, we've opened up the gates to far more experimentation. Whether you're limited by never being trained, not having physical access to lab, not having the time to poke around lab, or whatnot, you can now execute sophisticated biochemical and biophysical experiments in our Cloud Lab. Sure, anyone in theory can learn to run experiments and I love that there are home labs out there, but for many that's either impractical or just not the best use of their time. You must decide for yourself, are you happier spending several hours in the lab testing observations or those hours cranking up more agents to make new biology hypotheses?
Alas, we can't support every experiment, as much as we'd like to. Ginkgo Cloud Lab is purely data-in, data-out - you send us sequences, chemical ids, etc, and we send you back data. We're focused on workflows based on in vitro protein or RNA synthesis. I've seen cool proposals for projects that have safety or operational implications that mean we might refuse them; not a problem Amazon Cloud must deal with!
But that's glass half-empty view. Looking at it half-full, the field of experiments we can support is vast. Characterizing or engineering enzymes. Stringing together novel biochemical pathways in vitro. Testing engineered binders. Exploring biosensors. Given a set of basic Lego bricks you can't necessarily build photo-realistic buildings, but you can build a universe of structures!
So what are you going to build? Where does your bio-imagination take you? What computational prediction for you is just too exciting to leave as just a prediction?
Nice article. Checked your first paper: you published it back in 1994 and it is still behind paywall :-( Not a good situation to disseminate knowledge.
ReplyDeleteIn my defense, web journals didn’t exist when we published so the paywall concept (or open access) didn’t yet exist
DeleteBut fair point