DNNGIOR
in Software on Metabolic, Model
Highlights
We trained a deep neural network on >11k bacterial species to recover missing reactions
Reaction frequency and query similarity to the training data impacted performance
DNNGIOR models can simulate real data similar to CarveMe with fewer false positives
A data driven approach to gapfill Genome Scale Metabolic Reconstructions in an unbiased way.
Read more on DNNGIOR publication
