A strategy to gain a better understanding of disease proteins is to consider the interactions of these proteins by making use of the protein-protein interaction (PPI) data. It is known that proteins are composed of multiple functional domains. In this article, domain information is introduced to study two biomedical problems. For the first problem, a one to one domain-domain interaction (DDI) model is proposed to obtain specific sets of DDI for onco-proteins and tumor suppressor proteins respectively. Three specific sets of DDI, i.e. oncoprotein and oncoprotein, tumor suppressor protein and tumor suppressor protein, and oncoprotein and tumor suppressor protein, are derived from their PPI. Cross-validation test is conducted to benchmark the prediction sensitivity, specificity and F1 measure. It is found that the oncoprotein and cancer protein DDI set achieved a 74% and 84% F1 measure respectively. This indicates the feasibility of applying DDI model for cancer protein PPI studies, which can contribute to biomedical study Secondly, it is suggested that PPI, which is mediated by DDI, may be affected due to domain removal through the alternative splicing mechanism. Domain removal effects on liver cancer isoforms' PPI are studied. It is found that certain liver cancer-related isoforms mediated differential PPI. Information are integrated from three pieces of information, i.e. Gene Onotology (GO), gene expression level and PubMed, to provide supporting details for the findings. The Jaccard index is used to rank statistical significant DDI with the above combinations. Finally, web-based platforms have been set up to display the results of the two studied biomedical problems, i.e. http://ppi.bioinfo.asia.edu.tw/TsgOcgppi/ and http://ppi.bioinfo.asia.edu.tw/hccas/search.