Pathway Level And Metabolite Level Association Network Modelling And Its Application On CRC And GDM Metabonomics

In recent years,pathway analysis has been widely used in genomics,proteomics,and metabolomics.However,the traditional pathway analysis method ignores the correlation between the pathways,which makes it difficult to quantify the impact of the pathway on the whole network.Therefore,this paper proposes a pathway level association network(PLAN)modelling method and applies it to colorectal cancer(CRC)metabolomics.Moreover,this paper proposes a metabolite level association network(MLAN)modelling method and applies it to gestational diabetes mellitus(GDM)metabolomics.The main contents are as follows:1.Propose a PLAN network modeling method based on mutual information.The PLAN network uses pathways as nodes,evaluates the relationship between pathways with mutual information,and uses data processing inequality to reduce indirect connections between pathways.Moreover,the modeling method of differential PLAN is proposed to analyze the difference in metabolism between the experimental group and the control group.The restart random walk algorithm is introduced into PLAN and differential PLAN networks to evaluate the impact of a single metabolic pathway on the entire network.2.The PLAN network modeling method was applied to CRC serum metabonomics.The PLAN network and differential PLAN network of CRC are modeled to identify the key metabolic pathways,and the results are compared with those of the commonly used pathway analysis methods based on metabolite level network,namely ORA and NetGSA.The new method found some differential metabolic pathways in patients with CRC,such as alanine,aspartic acid and glutamic acid metabolism,tryptophan metabolism and phosphoinositide metabolism.Some recently published studies on CRC verify the accuracy of the differential pathways identified by PLAN.3.MLAN network modeling and its application on GDM metabonomics.This paper proposes a modeling method of differential MLAN network based on mutual information,and determines the screening criteria of differential metabolites.The serum samples of healthy pregnant women and GDM pregnant women were collected for NMR experiments.The NMR SpectFitting software developed by the laboratory was used to quantify the serum 1H NMR spectrum,and quantitative information of 47 metabolites was obtained.The differential MLAN network was modeled based on GDM data,and 16 differential metabolites were identified using the differential MLAN network.Then,based on the identified differential metabolites,pathway enrichment analysis was performed to identify the differential metabolic pathways of GDM,and compared with traditional methods.