ZeroFold: Protein-RNA Binding Affinity Predictions from Pre-Structural Embeddings

TL;DR

ZeroFold leverages pre-structural embeddings from foundation models to accurately predict protein-RNA binding affinity directly from sequences, effectively capturing conformational flexibility without requiring explicit structural data.

Contribution

This work introduces ZeroFold, a transformer-based model that uses pre-structural embeddings to predict binding affinity, addressing the challenge of RNA flexibility in structural biology.

Findings

ZeroFold achieves a Spearman correlation of 0.65 on the test set.

It outperforms existing structure-based and sequence-based predictors, especially with less similar training data.

Pre-structural embeddings effectively encode conformational ensemble information.

Abstract

The accurate prediction of protein-RNA binding affinity remains an unsolved problem in structural biology, limiting opportunities in understanding gene regulation and designing RNA-targeting therapeutics. A central obstacle is the structural flexibility of RNA, as, unlike proteins, RNA molecules exist as dynamic conformational ensembles. Thus, committing to a single predicted structure discards information relevant to binding. Here, we show that this obstacle can be addressed by extracting pre-structural embeddings, which are intermediate representations from a biomolecular foundation model captured before the structure decoding step. Pre-structural embeddings implicitly encode conformational ensemble information without requiring predicted structures. We build ZeroFold, a transformer-based model that combines pre-structural embeddings from Boltz-2 for both protein and RNA molecules through a cross-modal attention mechanism to predict binding affinity directly from sequence. To support training and evaluation, we construct PRADB, a curated dataset of 2,621 unique protein-RNA pairs with experimentally measured affinities drawn from four complementary databases. On a held-out test set constructed with 40% sequence identity thresholds, ZeroFold achieves a Spearman correlation of 0.65, a value approaching the ceiling imposed by experimental measurement noise. Under progressively fairer evaluation conditions that control for training-set overlap, ZeroFold compares favourably with respect to leading structure-based and leading sequence-based predictors, with the performance gap widening as sequence similarity to competitor training data is reduced. These results illustrate how pre-structural embeddings offer a representation strategy for flexible biomolecules, opening a route to affinity prediction for protein-RNA pairs for which no structural data exist.