It's a pity that I can't accept two answers as correct! Both of you guys helped a lot!
2 answers
Completely agree with the Completeness issue pointed by Michael. Still you could do some analysis with available data. I have done something similar to analyze hits from a genome scale RNAi screening.
Get all genes for a KO:Pathway in your list by parsing the ko.list
For example for the KO:Pathway path:ko0001 you can get all the components by parsing ko.list.
Now check how many ko components from your list are present.
You could assign a percentage to quantify this :
Completeness of pathway X = (No. of KO components in KO pathway X / No. of KO components in your list that belong to pathway X ) 100
This depends a bit on your definition of "see." ;-) For interactive exploration, I would recommend iPath which is based on KEGG data. There you can upload a list of genes and check the pathways by eye.
Computationally, you should be able to check if you have all the proteins for a pathway (e.g. if you have at least one hit for all the KO). But some pathways are more of a collection of independent modules, while in another pathway you might miss a crucial enzyme. So "completeness" is another difficult term.
I was also worrying about the "completeness" you mentioned... Is it possible to know (from a database maybe) if an enzyme is crucial or not for a pathway?
I guess you could do some graph-theoretic approaches to see if there is still "flow" through the pathway. However, you mentioned in your previous post that you're doing a metagenomic analysis, so it's always possible that the missing enzyme was not sequenced.
Log in to answer this question.