8F53 image
Entry Detail
PDB ID:
8F53
EMDB ID:
Keywords:
Title:
Top-down design of protein architectures with reinforcement learning
Biological Source:
Source Organism:
Host Organism:
PDB Version:
Deposition Date:
2022-11-11
Release Date:
2023-05-10
Method Details:
Experimental Method:
Resolution:
2.93 Å
Aggregation State:
PARTICLE
Reconstruction Method:
SINGLE PARTICLE
Macromolecular Entities
Polymer Type:polypeptide(L)
Description:RC_I_2
Chain IDs:A, B (auth: F), C (auth: K), D (auth: U), E (auth: J1), F (auth: Od), G (auth: Tc), H (auth: Ba), I (auth: B), J (auth: G), K (auth: L), L (auth: V), M (auth: Z), N (auth: Id), O (auth: Nh), P (auth: Sl), Q (auth: Xp), R (auth: Yb), S (auth: C), T (auth: H), U (auth: M), V (auth: W), W (auth: Ca), X (auth: He), Y (auth: Mi), Z (auth: Rm), AA (auth: Wq), BA (auth: Ae), CA (auth: D), DA (auth: I), EA (auth: N), FA (auth: X), GA (auth: Bb), HA (auth: Gf), IA (auth: Lj), JA (auth: Qn), KA (auth: Vr), LA (auth: Zt), MA (auth: E), NA (auth: J), OA (auth: O), PA (auth: Y), QA (auth: Dc), RA (auth: Fg), SA (auth: Kk), TA (auth: Po), UA (auth: Us), VA (auth: Hu), WA (auth: P), XA (auth: Gv), YA (auth: Q), ZA (auth: Fw), AB (auth: R), BB (auth: Ex), CB (auth: S), DB (auth: Dy), EB (auth: T), FB (auth: Cz), GB (auth: 0), HB (auth: 2)
Chain Length:54
Number of Molecules:60
Biological Source:synthetic construct
Ligand Molecules
Primary Citation
Top-down design of protein architectures with reinforcement learning.
Science 380 266 273 (2023)
PMID: 37079676 DOI: 10.1126/science.adf6591

Abstact

As a result of evolutionary selection, the subunits of naturally occurring protein assemblies often fit together with substantial shape complementarity to generate architectures optimal for function in a manner not achievable by current design approaches. We describe a "top-down" reinforcement learning-based design approach that solves this problem using Monte Carlo tree search to sample protein conformers in the context of an overall architecture and specified functional constraints. Cryo-electron microscopy structures of the designed disk-shaped nanopores and ultracompact icosahedra are very close to the computational models. The icosohedra enable very-high-density display of immunogens and signaling molecules, which potentiates vaccine response and angiogenesis induction. Our approach enables the top-down design of complex protein nanomaterials with desired system properties and demonstrates the power of reinforcement learning in protein design.

Legend

Protein

Chemical

Disease

Primary Citation of related structures