Extended Data Fig. 3: Schematic of the metabolic modeling approach used in our study.

A computational approach based on genome-scale metabolic modeling and constraint-based optimization was used to assist in the interpretation of Tn-seq data. To this end, a genome-scale metabolic model was used to simulate metabolism and perform in silico single gene deletion experiments on 66 different media compositions of varying complexity (that is, single carbon media, double carbon source media, and rich media). Media selection was guided by the assumed mucus composition and Tn-seq data. Dimensionality reduction techniques and clustering were then used to compare the in silico data with the Tn-seq data to infer carbon utilization on the mucosal surface. Abbreviations: LB, Luria-Bertani broth; CF, cystic fibrosis; SCFM, synthetic cystic fibrosis sputum medium.