iiinova — where your digital self keeps evolving

Kai Yi

@kaiyi94.bsky.social

69 followers 76 following 10 posts

Machine Learning, Protein Design. Postdoc in Sjors Scheres Lab at MRC-LMB

Posts Replies Media

Reposted by Kai Yi

Sjors Scheres @sjorsscheres.bsky.social · 30/09/2026

Our adventures into #proteinfolding: GuideFlip designs flexible molecular interactions by discrete flow matching to design sequence & structure together! Grateful for funding by Love Tito's (@titosvodka.bsky.social) and the @ukri.org. 🥳

02513

Reposted by Kai Yi

Kiarash Jamali @kjamali.bsky.social · 29/09/2026

Please check out our cool new method, GuideFlip! Optimised for binders against disordered proteins

0113

Kai Yi @kaiyi94.bsky.social · 29/09/2026

7/7 Huge thanks to my co-authors @qcchen.bsky.social , Dongqi Zhang, Pengfei Tian, Jane L. Wagstaff, Stephen H. McLaughlin, Christopher G. Tate, @kjamali.bsky.social and @sjorsscheres.bsky.social, and everyone who supported this work! Preprint: www.biorxiv.org/content/10.6...

biorxiv.org

De novo design of flexible protein interactions with GuideFlip

De novo design of protein binders requires a target structure. However, for flexible targets, such as intrinsically disordered proteins, this structure does not exist until the binder has stabilized t...

040

Kai Yi @kaiyi94.bsky.social · 29/09/2026

6/7 We further apply GuideFlip to a curated set of 285 human DisProt targets. For 177 (62.1%), at least one design passed our computational filters. These are predicted candidates. Explore complexes and download sequences: huggingface.co/spaces/ykiii...

121

Kai Yi @kaiyi94.bsky.social · 29/09/2026

5/7 De novo design of nanobody CDRs against β1AR yielded 12 hits from 16 selected designs (75%), with KD as low as 13 nM. A designed nanobody stabilized the active state. Its cryo-EM structure closely matched the design.

110

Kai Yi @kaiyi94.bsky.social · 29/09/2026

4/7 For RBX1’s disordered N-terminus, GuideFlip gave 5/12 hits across two rounds (41.7%). After 2nd place in the GEM × Adaptyv competition, follow-up designs reached KD = 4.5 nM. Thanks to GEM and @adaptyv.bio for organizing, and @adaptyv.bio for testing!

100

Kai Yi @kaiyi94.bsky.social · 29/09/2026

3/7 GuideFlip designs binders to the fully disordered C-terminus of α-synuclein, with our strongest binder at KD = 211 nM. SPR, truncation experiments, mutagenesis and solution NMR support recognition of the intended region.

110

Kai Yi @kaiyi94.bsky.social · 29/09/2026

2/7 GuideFlip combines an all-atom inverse-folding prior (ADFlip) with gradients from AlphaFold2. Binder residues are assigned progressively as the complex is re-predicted, allowing sequence and the binding interface to evolve together.

110

Kai Yi @kaiyi94.bsky.social · 29/09/2026

1/7 Excited to share GuideFlip! We use guided discrete flow matching to co-design protein sequences and structures for flexible interactions, with experimental validation across three systems: α-synuclein, RBX1 and active-state GPCR nanobodies.

13527

Kai Yi @kaiyi94.bsky.social · 28/04/2026

Thanks @adaptyv.bio organizing the interesting experiment and congrats to other participants! proteinbase.com/collections/...

proteinbase.com

GEM x Adaptyv: RBX1 Binder Design Competition - Results

Results from the GEM x Adaptyv RBX1 Binder Design Competition.

020

Kai Yi @kaiyi94.bsky.social · 28/04/2026

Thrilled with the our results with @kjamali.bsky.social and @sjorsscheres.bsky.social from the recent binder competition! 🎉 2 of our 7 designs successfully bound to the target, with affinities ranking 2nd and 3rd overall. All of these designs targeted the disordered region of the protein.

1216

Kai Yi @kaiyi94.bsky.social · 14/07/2025

Excited to share our paper at #ICML: All-atom inverse protein folding through discrete flow matching with @kjamali.bsky.social and @sjorsscheres.bsky.social : openreview.net/forum?id=8tQdw…. If you are at ICML, let’s connect and talk generative models&protein design!

openreview.net

OpenReview

Promoting openness in scientific communication and the peer-review process

230