#37 · Structural bio & protein design

Computational prediction of ribozymes, communication residues, and DDX3X unwinding

The common thread is computational prediction paired with an explicit check against validation data. Discrete-time quantum walks with Shannon entropy on protein contact networks identify residues involved in long-range interactions, validated on AsLOV2. AbRiDGed, an ab initio method, predicts millions of self-cleaving ribozymes across species and identifies roughly 150 novel classes. The DDX3X study, combining coarse-grained simulations with deep learning, describes ATP-independent, entropy-driven dsRNA unwinding through rare strand-displacing intermediates. The validation orientation is dominant rather than universal, since the DDX3X work is primarily mechanistic. What is new is the use of network-entropy walks to localize communication residues and genome-scale ab initio discovery of catalytic RNAs.