Among them, 2260 genes were KO target genes (knockout set).
The number of parent knockout sets, P, was determined heuristically.
Then, top P knockout sets with regard to this score were chosen for the parent knockout sets.
Knockout set: the set of reactions globally affected by the knockout of a given reaction.
If of a selected knockout set became positive or zero, this knockout set was excluded from the iterative screening and was stored as a candidate knockout set for further analysis.
In Algorithm 1, we perform a multiple knockout analysis with a fixed knockout set.
For each generation, all possible single reaction knockouts were further added to the parent sets (leading P × N knockout sets), and among these knockout sets that increased from the parent sets, P knockout sets with the largest were selected as the parent sets of the next generation.
These knockout sets contained 25 unique reactions that were selected for the OptKnock screening (Supplementary Table S3).
Here, we analyzed the similarity of knockout candidate sets among the target metabolites, with the aim of identifying a common knockout set for the production of a specific class of metabolites.
On the one hand, complete methods have been proposed, which systematically explore all possible knockout sets up to a given size, e.g. [ 9, 10].
