9 replies [Last post]
Susume's picture
User offline. Last seen 1 day 21 hours ago. Offline
Joined: 10/02/2011

New paper (preprint) from the Baker lab about successful denovo design of protein binders for covid-19 spike protein: https://www.biorxiv.org/content/10.1101/2020.08.03.234914v1

Also a nice summary and explanation of the paper: https://www.news-medical.net/news/20200804/Picomolar-inhibitors-to-SARS-CoV-2-proteins.aspx

Susume's picture
User offline. Last seen 1 day 21 hours ago. Offline
Joined: 10/02/2011
Similar results from Neoleukin Therapeutics

NeoLeukin Therapeutics in Seattle (which I think is a Baker lab spin-off company) has a similar preprint out, but entirely different author list - apparently they have been working independently on the same goal and have a different set of binder designs ready to publish.

Summary and explanation of the Neoleukin paper:

georg137's picture
User offline. Last seen 13 hours 25 min ago. Offline
Joined: 08/07/2010
Groups: Contenders
Are scientists embarrassed by Foldit connection?

The protein twitter feed is quite positive about the pre-print, and very complimentary about Baker Lab. Almost all of the co-authors are associated with the Institute for Protein Design and the Department of Biochemistry at the University of Washington in Seattle, the primary sponsor of Foldit. The paper features several diagrams of many triple alpha helices binding to corona virus spike surfaces. The virus and binder design configuration space appears to be the same space used in the in-silico work-products that Foldit teams and soloists have been designing, refining, and submitting to scientists since the start of the pandemic. Looking at the talent and qualifications of the investigators, and the track record of the Foldit contributor community, it is not mere coincidence that the triple helix strategy should arise spontaneously in their independent explorations. While it may be technically correct not to formally go into the strong similarities to the Foldit COVID-19 binder configurations, the authors may have overlooked an opportunity to bolster their findings by at least mentioning the work of their citizen-science counterparts. They definitely missed an opportunity to align with long-term seasoned Foldit contributors, their locked-in allies in explosively competitive work.


bkoep's picture
User offline. Last seen 14 hours 7 min ago. Offline
Joined: 11/15/2012
Groups: Foldit Staff
Response to georg137

Not at all embarrassed!

The results in both of these preprints are a feat of exceptional effort, and these author lists represent the world's leading experts on protein binder design and coronavirus spike proteins. That is not hyperbole -- there is nobody in the world better qualified to tackle these problems. And some of these scientists have been helping the Foldit team directly, so that we can add their binder design tools to Foldit.

Rest assured that Foldit had absolutely nothing to do with the designs in either paper, and we shouldn't read too far into any supposed similarities. (I believe some of the earliest Baker Lab designs were ordered for testing before the first Foldit puzzle even closed.) The prevalence of 3-helix bundles is not a coincidence; it is well established that helical bundle designs have high success rates, and they are the workhorse of most protein design applications. (The first protein designs in the 1980s and 90s were helical bundles, and the prevalence of helical bundles is clearly evident in modern high-throughput experiments.)

Foldit is not built for binder design (yet!). The Foldit monomer design paper came out only last year. Our original plan was to begin binder design puzzles much later in 2020, once all of the binder metrics and other tools were in place. But when the coronavirus spike structure was released in February, we jumped in with what we had. We've been more than impressed by Foldit players' work and willingness to adapt, as well as your patience while the Foldit team works on improvements.

I also want to stress that this does not mean the Foldit coronavirus binder puzzles have been wasted effort. On the contrary, these puzzles have laid the groundwork for protein binder design in Foldit. We now have a very clear picture of the challenges facing binder design in Foldit, and a solid baseline for future experiments. Scientifically, there is still plenty to be learned about protein binder design, and we look forward to challenging Foldit players with harder targets (there are binder targets out there that even these experts won't touch!).

In short, we should not be discouraged that the world's top minds (working nonstop with virtually unlimited resources) were able to more quickly solve a problem that Foldit is not yet optimized to solve. Stay tuned, we may dig into these preprint results in a future blog post.

Joined: 09/29/2016
Groups: Gargleblasters
Love the info provided there!

And this also highlights something that I absolutely love about science, which if you'll permit me to kludge the original quote since I can't quite recall it:
"There is no such thing as failure in science..." (or bad data, incorrect results)
"If you've attempted to prove or solve something 500 times before your finally succeeded, you did not fail those initial 500 times, you simply learned 500 ways NOT to accomplish it!"

Granted, this is true about more things in life than just science, but it does ring especially true there. All the times an experiment is run and doesn't yield the expected results (or what have you), that's still data! It's really only a "failure" if you do not look at that data and learn something from it, or use it to improve things to eventually reach your objective. Sometimes to be dis-proven is just as much of an achievement as it is to be proven.

However, even if the conclusion you've come to ends up disproving what your were trying to accomplish... Well? That's still a data point in the grand scheme of it all, and therefore still useful in the end, as it checks off something off on a list somewhere.

TL;DR - All of our results ultimately lead to Foldit getting better, and as such, that means our efforts have not been in vain! :D

jeff101's picture
User offline. Last seen 9 hours 13 min ago. Offline
Joined: 04/20/2012
Groups: Go Science
Scene from the movie "iOrigins":

shows a scene from the movie
"iOrigins" reflecting the same idea.

Also, I think "The Cat in the Hat"
had a search strategy in which he'd
mark with an X every spot he'd
already checked.

beta_helix's picture
User offline. Last seen 13 hours 28 min ago. Offline
Joined: 05/09/2008
Groups: None
Full Science pdf
georg137's picture
User offline. Last seen 13 hours 25 min ago. Offline
Joined: 08/07/2010
Groups: Contenders
IPD is on the cutting edge

IPD is on the cutting edge of anti-virals. Congratulations to LongXing Cao, David Baker, et. al. The results are a COVID breakthrough, and a bioinformatics coup. Well done.

georg137's picture
User offline. Last seen 13 hours 25 min ago. Offline
Joined: 08/07/2010
Groups: Contenders
More on COVID-19

For those interested in broader COVID aspects, the link below goes to many COVID-19 research papers, articles, preprints, letters and commentaries at the CelPress Coronavirus Resource Hub.


Susume's picture
User offline. Last seen 1 day 21 hours ago. Offline
Joined: 10/02/2011
New York Times writeup

Nice writeup in NYT about the mini binders from Baker lab and Neoleukin: https://www.nytimes.com/2020/11/21/science/coronavirus-antibodies-artificial-intelligence.html

(This is part of their covid coverage, so it is not behind a paywall.)


Developed by: UW Center for Game Science, UW Institute for Protein Design, Northeastern University, Vanderbilt University Meiler Lab, UC Davis
Supported by: DARPA, NSF, NIH, HHMI, Amazon, Microsoft, Adobe, Boehringer Ingelheim, RosettaCommons