FIG 6.

Diagrams of how deep curation could be used to evaluate the cross-consistency of heterogenous data sets. (A) In the building of the M. genitalium whole-cell model, it was noted that the mass of the total DNA could be calculated using two different methods, one from the measured mass fraction of DNA and the other from the genome sequence, and was significantly different from each other. This led to a suggestion that the measurement for the mass fraction of DNA contained errors. (B) In the E. coli whole-cell model, raw data from various sources were tied together into a unified, mechanistic model, using multiple layers of curation. Certain outputs from the resulting model, however, were inconsistent with reported experimental values, which led us to reevaluate the data sources, parameters, and equations used in the model to identify the source of the discrepancy. Using a set of criteria, including the reproducibility of results reported in the literature, we concluded that the mRNA expression data underestimated the expression of genes that encode RNA polymerase and ribosomal subunits.