Davis's lab develops CryoDRGN for ribosome cryo-EM analysis
Updated
Updated · MIT News · May 5
Davis's lab develops CryoDRGN for ribosome cryo-EM analysis
4 articles · Updated · MIT News · May 5
At MIT, Associate Professor Joey Davis's team said the 2021 neural-network method reconstructs full ensembles from hundreds of thousands of frozen-particle images.
The approach showed blocked ribosome-assembly steps can produce many alternative structures, suggesting cells build the protein-making machines through flexible, non-linear pathways rather than a fixed sequence.
Davis now aims to scale cryo-EM throughput to generate structural datasets that could improve AI protein-prediction models, especially for poorly predicted regions of sequence space.
With AI now designing proteins and decoding life’s machinery, are we on the verge of a new biological industrial revolution?
We thought life’s protein factories followed a strict blueprint. What if they improvise, and what does this mean for disease?
A new theory casts ribosomes as 'selfish' cellular masters. How does this upend our entire understanding of evolution and life?
Modeling Continuous Structural Heterogeneity in Ribosome Assembly with CryoDRGN and MAVEn
Overview
CryoDRGN revolutionizes cryo-EM analysis by using a neural network with a variational autoencoder to model continuous structural heterogeneity, overcoming limitations of traditional discrete classification. This approach enables pose-free ab initio reconstruction, capturing dynamic molecular conformations. Ribosome assembly is shown to be a flexible, non-linear process with multiple pathways and diverse quality control mechanisms, reflecting evolutionary adaptations. Tools like MAVEn reveal how KsgA proofreads immature ribosomal subunits by remodeling key RNA structures. Advances in high-throughput cryo-EM combined with AI structure prediction create a powerful synergy that addresses each method's limitations, accelerating drug design by revealing transient states. Despite challenges like computational cost, ongoing efforts aim to enhance accessibility and integrate these technologies for deeper biological insights.