Christopher Frank
@chrisfrank662
Followers
813
Following
389
Media
48
Statuses
199
De novo Protein Designer DNA Origami wizzard I like lifting rocks PhD student @TU_Muenchen
Hallbergmoos
Joined June 2013
Finally, out in @ScienceMagazine today, our work on de novo protein design. Together with @HattoriLab , @sokrypton & @hendrik_dietz we showed how we can drastically improve AF2 based hallucination, surpassing, or reaching SOTA diffusion models.
9
130
571
🤠.
0
0
3
Thats how one does ads 😅.
Ski fast…and sequence faster. Plasmidsaurus now lives in Germany! 🇩🇪. Overnight sequencing for ALL of Europe.Pickup ~4PM → Results 8AM. 🦖 Dropboxes in 14 European countries. and the dino is looking for MORE! Request your free dropbox here: Sequence
0
1
4
RT @woodyahern: New enzymes can unlock chemistry we never had access to before. Here, we introduce RFdiffusion2 (RFD2), a generative model….
0
75
0
I became aware of the true Munich-Augsburg rivalry today🚀
Dear rocket colleagues from the neighboring city, we send our very best wishes for your launch campaign of Spectrum! 🚀. We know very well what an incredible challenge it is to develop, build, test and finally launch a rocket. And how much passion, perseverance, nerves of steel.
0
0
1
RT @e_estonia: Estonia is now 0% bureaucrazy!.We’re proud to announce that every government service in Estonia is now fully digital. Whethe….
0
194
0
RT @miangoar: 🧵1/8 In ~2020, diffusion models gained popularity with famous AIs like DALL-E. But diffusion models work well for protein des….
0
19
0
0
16
0
I somebody up to date about the legal aspect of redesigning existing antibodies etc? Just wondering how dis-similar a sequence needs to be, to be able to be used again? .Nevertheless, the results are 🔥🚀.
PROTEIN DESIGN COMPETITION ROUND 2. The long awaited results are here -- and they don’t disappoint! . 📈 We doubled the number of proteins we tested in our lab from 200 to 400! . 🧬 Out of those 400 proteins, 378 expressed (95% expression rate!). 🚀 Out of those 378 expressed
7
2
28
For round three we need a @Foldit puzzle. Would be hilarious if some random person outperforms all the ML methods 😅.
Our 2nd protein design competition saw designers racing to optimize their protein scores until the last minute. Want to know which were the most successful ML methods? Was RFdiffusion more popular than Bindcraft? . Check out our analysis of the submissions in our new blog post
5
3
21
This is awesome! Generate a binder, insert it onto a CAR, kill cancer cells reliably! Stunning results!.
Cancer escapes immunotherapies by mutating its surface. This means even groundbreaking #CARTCell therapies fail too often because of cancer resistance. But what if we could use #GenerativeAI to design CAR-T with de novo binders faster and smarter, targeting even resistant tumors?
0
0
8
RT @BiologyAIDaily: Alphafold2 refinement improves designability of large de novo proteins. • Introducing af2cycler, a pipeline that enhanc….
0
25
0
If you have questions, suggestions etc, please let us know! We are still trying to make things better, get more data, controls and expand usefulness and obviously – if you wanna look at one if this letters in cryo: we are more than happy to help you out! :).
3
1
5
This was a super fun collaboration with @sokrypton & @hendrik_dietz . Thanks to @DominikSchiwie0 for all the help with designing shapes and twitterless Lara Fuß for all the help with colony picking.
1
0
1
Now, where is the limit? We went even further, testing 2000 AA designs in silico, but up to this point we haven’t succeeded in finding a designable sequence for our protein pretzel :D
1
0
1
We tested just one sequence, which worked out of the box, and we got clean samples right after the Nickle-NTA purification. TEM analysis also showed that the designed protein seems to agree very well with the designed shape.
1
0
1
We then wanted to push the size limits even further. When designing a 1500 AA triangle shaped protein we noticed that neither ESMFold, nor AF2 with IG & AA could predict the sequences confidently, so we went with AF3, which did a better job.
1
0
1
I was honestly surprised how well this worked. The TEM micrographs are so cool :D Just a bunch of tiny shapes
1
0
2