Open Access

Christian Gebhardt, Pascal Bawidamann, Anna-Katharina Spring, Robin Schenk, Konstantin Schütze, Gabriel G. Moya Muñoz, Nicolas D. Wendler, Douglas A. Griffith, Jan Lipfert, Thorben Cordes

• Attaching fluorescent dyes to biomolecules is essential for assays in biology, biochemistry, biophysics, biomedicine and imaging. A systematic approach for the selection of suitable labeling sites in macromolecules, particularly proteins, is missing. We present a quantitative strategy to identify such protein residues using a naïve Bayes classifier. Analysis of >100 proteins with ~400 successfully labeled residues allows to identify four parameters, which can rank residues via a single metric (the label score). The approach is tested and benchmarked by inspection of literature data and experiments on the expression level, degree of labelling, and success in FRET assays of different bacterial substrate binding proteins. With the paper, we provide a python package and webserver (https://labelizer.bio.lmu.de/), that performs an analysis of a pdb-structure (or model), label score calculation, and FRET assay scoring. The approach can facilitate to build up a central open-access database toAttaching fluorescent dyes to biomolecules is essential for assays in biology, biochemistry, biophysics, biomedicine and imaging. A systematic approach for the selection of suitable labeling sites in macromolecules, particularly proteins, is missing. We present a quantitative strategy to identify such protein residues using a naïve Bayes classifier. Analysis of >100 proteins with ~400 successfully labeled residues allows to identify four parameters, which can rank residues via a single metric (the label score). The approach is tested and benchmarked by inspection of literature data and experiments on the expression level, degree of labelling, and success in FRET assays of different bacterial substrate binding proteins. With the paper, we provide a python package and webserver (https://labelizer.bio.lmu.de/), that performs an analysis of a pdb-structure (or model), label score calculation, and FRET assay scoring. The approach can facilitate to build up a central open-access database to continuously refine the label-site selection in proteins.…