Predicting of Autism Spectrum Disorder using Gene Ontology

A massive amount of genomic data is available today. Identifying the disease genes from this vast data becomes a big challenge in the field of bioinformatics. Therefore, perceiving the relationship between disease causing genes is essential for developing new treatment and better clinical diagnostics. Moreover, the biggest challenge when dealing with complex diseases. Their genes are highly heterogeneous, which makes the process of disease gene identification more difficult. Machine learning classification techniques are utilized to predict disease genes using Gene Ontology (GO). Annotating genes with gene ontology terms from GO help to measure the semantic relationships between them. The proposed classification approach is assessed using the candidate genes of Autism Spectrum Disorder (ASD). Different gene functional similarity measures were applied using different classifiers to dig out the semantic similarity between ASD genes. The performance of the approach is analyzed compared to other techniques, which are based on gene expression and protein-to-protein network. It records 73% to 80% using Random Forest classifier for predicting new ASD genes.