Sophisticated mass spectrometry (MS) techniques in conjunction with covalently-labeled protein residues yield exposure information useful in protein structure prediction. However, easy and reliable translation of this information into accurate structural models remains particularly challenging, due to a lack of standardized and automated methods. We developed and validated a software tool, termed “MS-Fold”, that allows data generated from covalent labeling MS studies (e.g. solvent-exposed protein residues) to be effectively used to guide protein structure prediction algorithms. First, we developed a script to convert the raw MS labeling data into a standardized format that contains quantitative residue exposure information. This information describes the solvent exposure of labeled residues and constitutes a useful restraint for computational protein structure prediction. Second, we developed a custom MS-scoring function that favorably scores exposure of labeled residues in protein models. Finally, these two functionalities were implemented into the Rosetta protein structure prediction software. To evaluate the effectiveness of MS-Fold we benchmarked its performance using three proteins of known structure for which we have obtained experimental covalent labeling data. For all benchmark proteins, models with lower average RMSDs (i.e. better model quality) were built when MS scores were included. Models with less than 3 Å RMSD to the native structure were built for all proteins. MS-Fold proved to be a powerful hybrid method allowing for covalent labeling MS data to be used in protein structure prediction. We are also working to incorporate Surface Induced Dissociation (SID) data in protein complex prediction.

Protein structure prediction guided by covalent labeling mass spectrometry data

254th National Meeting (2017)

The official archive and repository for all ACS Meetings & Expositions abstracts and select posters or presentations.