Candidate gene prioritization by network analysis of differential expression using machine learning approaches