Targeted experimentation to increase functional coverage in omics dataset. Escherichia coli BW25113

To increase the GO coverage of Ecomics, we investigated which new conditions would lead to the most informative transcriptional profiles. We first transcriptionally profiled and included in the compendium 9 KO experiments that we had identified before as being highly informative for both GO coverage and model performance. We assumed that a GO term is represented in the compendium if the compendium has profiles where one or more genes that include that GO term have been perturbed. We identified the gene rank iteratively under the assumption that all genes of higher rank have been perturbed. Based on the resulting ranked list, we performed transcriptional profiling for the top 16 genetic perturbations (in triplicate; 48 profiles total), which led to an increase in coverage by 19.8% for KEGG, 24.3% for BP and 63.9% for MF. Overall design: Genome-wide expression profiling of 25 single-gene knockouts for BW25113 and its wild-type grown with either LB or M9 media supplemented with glucose (total 87 samples).