Fig. 1

Schematic of modeling workflow and the hypothetical antioxidative properties of a protein. a The genomes of 1764 strains of E. coli were gathered and orthologous genes were mapped to the reference E. coli K-12 MG1655 strain. External data sources were integrated to gather protein sequence and structure annotations with regards to susceptibility of oxidative damage, such as the locations of metal-binding sites [38], known carbonylation sites [39], and known cysteine damage sites [40]. The structural proteome is further categorized into protein groups by their annotated localization and functionality (see Additional file 1: Table S1). We conducted gene deletions upon the genome-scale metabolic model of MG1655 integrated with ROS generating reactions (iML1515-ROS [41, 42]) for strain-specific predictions of basal ROS production levels, defining a strain’s “ROStype”. We utilized the GEM-PRO pipeline [43, 44] to select representative protein structures for 95% of the MG1655 proteome. b A hypothetical protein resistant to oxidative damage. Protein sequence and structure properties are highlighted based on previous studies finding enrichment of these properties in aerobes or long-living organisms. Structural properties defined by locations in 3D space, such as surface-exposed residues or those in a specified radius within a metal-binding site, are used to further divide a single protein into residue groups (see Additional file 1: Table S3)