Neural spectroscopy of AlphaFold2 reveals encoded protein conformational landscapes
Machine Learning
Summary
The authors studied AlphaFold2's internal parameters, which are usually only used to turn protein sequences into structures. They found that by smoothing and scaling these parameters, the model reveals patterns of how proteins might change shape, reflecting what experiments have shown about protein folding. This means AlphaFold2 has learned more than just a single structure—it encodes some knowledge of protein flexibility and alternative forms. The authors call their analysis method "neural spectroscopy," showing that AlphaFold2's training lets it capture structural information beyond static predictions.
Authors
Kaustav Mehta
Abstract
AlphaFold2's 93 million parameters, shaped by the evolutionary record of protein structure encoded in the Protein Data Bank and in sequence alignments, are conventionally treated only as machinery for converting sequence to structure. We propose they are also a scientific object that can be analyzed directly: a learned encoding of protein conformational organization that can be probed and characterized. By smoothing the Evoformer's weight tensors with a Gaussian convolution and scaling the result, we show that the trained model produces physically structured conformational landscapes. Under perturbation, ubiquitin's native contacts break in the order established by decades of folding experiments. For KaiB, five independently trained models agree that the alternative fold is not recovered under perturbation. For alpha-synuclein, five models produce five different but coherent landscapes, mapping where the training signal has determined the representation and where it has not. Matched-power noise controls confirm that random corruption of equal magnitude produces debris, not conformations. The model learned to predict static structures; the conformational organization visible under perturbation was not an explicit training target, suggesting it emerged as a byproduct of that objective. AlphaFold2's weights appear to encode structural constraints, shaped by evolutionary and structural training data, that extend beyond what unperturbed inference reveals. We call the approach of reading them neural spectroscopy, and Scaled Gaussian Convolution one such protocol.