Structure guided significance testing correction for hydrogen deuterium exchange mass spectrometry

Abstract

Hydrogen deuterium exchange mass spectrometry (HDX-MS) is a powerful technique to probe changes in protein structural dynamics. In differential settings, HDX-MS compares dynamics between protein states, such as conformational changes resulting from antibody-antigen binding or the effects of mutations. As the method becomes more high-throughput, the number of comparisons between peptides and states grows, creating a multiple hypothesis testing challenge where some observed changes may result from statistical randomness rather than biological differences. While this problem can be addressed by applying multiple hypothesis testing correction techniques like false discovery rate (FDR) control, current methods assume statistical independence - an assumption violated by peptide overlap and the influence of protein structure. Here, we develop a structural false discovery rate (sFDR) that accounts for these dependencies by integrating sequence and structural information to estimate the effective number of independent tests. Our approach significantly improves statistical power to detect genuine changes in protein dynamics measured by HDX-MS, as demonstrated through extensive validation using both simulated and experimental datasets. The sFDR method maintains robustness to structural uncertainty, making it applicable even when only predicted structures are available. This advancement enables more sensitive detection of conformational changes in challenging systems with subtle dynamic differences and reduces the number of replicates needed to obtain confidence in results. The method is easily accessible through a Google Collab Notebook and can be completed in minutes.