Finding phenotype related pathways via biological networks comparison

Abstract



Why some species (or strains of a species) exhibit certain phenotypes (e.g. aerobic, anaerobic, pathogenic etc.) while the others do not is an important question to be answered. Apart from the conventional genomic study, studying the metabolism of the two groups of species may discover the corresponding pathways that are conserved in one group but not in the other. However, only a few tools provide functions to compare two groups of metabolic networks which are usually limited to the reaction level, not the pathway level.

In this dissertation, a problem named DMP (Differentiating Metabolic Pathway) problem was formed. Given two groups of metabolic networks, it aims at finding conserved pathways exist in first group, but not the second group. The problem also captures the mutation in similar pathways and derives a measurement (p-value and e-score) for evaluating the significance of the pathways.

An algorithm, DMPFinder, was developed to solve the DMP problem. Experimental results show that DMPFinder is able to identify pathways that are critical for the first group to exhibit a certain phenotype which is absent in the other group. Some of these pathways cannot be identified by other tools which only consider reaction level or do not take into account possible mutations among species.