Fig. 2

TraDIS - Upset plot comparing the regression results to Goodall et al. paper, Keio [18] and PEC [19] databases. We compared the list of essential genes resulting from our reproduction of the Goodall and al. study with their results and those identified by two external databases Keio and PEC. We can see that our replication predicts slightly more essential genes than the manually curated results provides, 364 genes against 353. Out of the 232 core essential genes identified in the study, we successfully identified 230. We identified most of the genes predicted by the paper that are absent in PEC or Keio (101 genes). Our replication identifies 13 essential genes identified in PEC but not in the original study in addition to the 35 genes identified everywhere but in Keio. The genes predicted by only one of the sources are of similar magnitude, a dozen genes each. They can be explained by manual annotation, parameter change, and domain essential genes