Breakdown of the false predictions by the source of false prediction. The reasons for false predictions are as follows: transcriptional regulation (regulation), model structure, accumulation of toxic intermediate in vivo (accumulation), dead ends in the model (dead end), discrepancy in the experimental data (exp discrepancy), gene–protein reaction associations (isozyme), unknown, in silico media composition (media), in silico biomass composition (biomass), and other cellular processes not included in the model (other). Results are shown for each experimental condition, including essential genes (essential) and slow growth genes (slow) on rich media. In addition, the distributions of the sources of false predictions are shown for false-positive (FP), false-negative (FN), and unique false predictions (unique) separately.