Text-Mining and Visualization Approach Help Interpret Experimental Data and Make Hypothesis

With the continuously grow of scientific literature, it becomes difficult for biologists to search for information by hand to better understand specific genes. When get a list of genes, which may play important roles in their experiment systems, either up regulated or down regulated, it’s difficult for scientists to get an over all idea and interpret this list. In our first study, we present how linking genes with other genes sharing iTerms for more than a certain proportion can reveal which pathways are more affected by these genes. We chose AKT, MTOR, PTEN for this study. Each gene have a certain number of iTerms, which are informative terms retrieved based on frequency of occurrence in the gene’s literatures by a web-based tool called eGIFT. Genes from same pathway or having similar functions share common iTerms. To retrieve a list of AKT,PTEN, MTOR related genes, we retrieved genes that share more than a certain proportion of iTerms with those iTerms associated with AKT, PTEN and mTOR. Subsequently, pathways are ranked according to the frequency of genes from this list that were mapped to, we successfully identified NGF signaling pathway; FGFR signaling pathway; PDGF signaling pathway are highly affected by these three genes. We have extended this study to analyze a transcriptome analysis of chicken retina. Our approach shows that combining multiple text-mining resources allows biologists to rapidly interpret data and develop new hypotheses.