KEGG automated metabolite protein interaction network for graph-model (KAMPING)

Introduction

KEGG features five types of relations: PPrel, GErel, PCrel, ECrel, and maplink. The following figure shows the relation types and their corresponding descriptions.

Of the five relation types, ECrel and PCrel describe protein-metabolite interactions. The two entries of ECrel are two protein (enzyme) entries, with the value of the relation being the metabolite entry, it can be glycan or compound (e.g. cpd:C05378 gl:G00037).

entry1    entry2	type	value	name
hsa:130589	hsa:2538	ECrel	cpd:C00267-90	compound

The first entry of PCrel is a compound entry, and the second entry is a protein entry. The name and value of the relation represent the effect of this compound on the protein. The name can be activation, inhibition.

entry1    entry2	type	value	name
cpd:C15493-60	hsa:6258	PCrel,PCrel	-->,+p	activation,phosphorylation

Due to data parsing, there can be more than one relation between two entries. For example, the following entry has two the value and name, the value and name are separated by a comma.

Metabolite-protein interaction relation:

We can process ECrel relation by expanding it into two binary relation (A-B), also called SIF (simple interaction format) in BioPAX standard, with first relation with original entry1 as the new entry1 and metabolite as the new entry2 in the first new relation. Likewise, the second new relation has the original entry2 as the new entry1 and the metabolite as the new entry2.

entry1    entry2	type	value	name
hsa:130589  cpd:C00267-90 ECrel compound compound
hsa:2538 cpd:C00267-90  ECrel compound compound
# todo: havn't decide the value and name after expanding

Code

After retrieve all relation in an kegg pathway

knext mpi --input data/kegg/hsa-ecrel-expanded.txt --output data/kegg/hsa-ecrel-expanded-mpi.txt