Draft genome & pathologic
What do you think is the best practice when reconstructing a metabolic network using pathwaytools/pathologic and having only a draft genome?
Draft genomes can have a lot of unconnected contigs. Pathologic needs a annotation file as input but the annotation of draft genomes is also not always trustworthy.
I'm interested in this because I encountered different results of predicted reactions and pathways when I created two pgdbs out of the same genome but using two distinct annotation files done by different prokka versions. I know that the pathway prediction algorithm is doing some kind of text mining with the annotation file but it is also using the raw sequence data to annotate gene function itself.
What is the balance between text mining and redoing the annotation? Can this be changed for draft genomes? Do you have recommendation for annotation software which works best with pathologic? (prokka, rast, ... ?)