Identification of candidate genes for Rituximab response in rheumatoid arthritis with weighted gene co-expression network analysis

Some rheumatoid arthritis patients are not clinical response to Rituximab treatment, so the identification of candidate genes for Rituximab response is very important which might contribute to improve the response chance. This paper identifies the candidate genes in a module view, which is more robust and biologically meaningful compared to the conventional gene differentially expressed analysis. The Weighted Gene Co-expression Network Analysis algorithm is firstly used to construct the gene co-expression network of rheumatoid arthritis. The co-expression network is then divided into several gene modules that might correspond to some biological functions with the dynamic tree cut algorithm. Gene expression profile based method and module eigen-genes based method are proposed to calculate the significance of gene modules that respond to Rituximab treatment. The experiment is conducted on a real dataset (GSE54629) from GEO, on which the co-expression network is constructed. The network finally is separated into six gene modules (blue, brown, green, grey, turquoise, yellow), of which the green module that consists of 61 genes is consistently confirmed most significant by both of the aforementioned two methods. The encouraging experimental results in comparison with several other representative papers demonstrate the effectiveness of our method to identify the candidate genes for Rituximab response.