Protein domains are compact regions that usually carry out discrete biological function and could fold independently. While in instances of low mutation rates, it is quite obvious to recognise protein domains from mere sequence information, recognising domain sequences within genomes become challenging at high evolutionary divergence. Recognition of protein domains in whole genomes using computational approaches provide valuable insights into such protein families within and across genomes. In the human genome, in particular, additional analyses and structural rationale of disease-causing mutations can be performed. In my talk, I will describe our attempts on all these directions – our computational approaches to recognise domains at high evolutionary divergence, search in whole genomes and across genomes and mapping mutations on structural domains. The relevance of such studies on biological problems will be illustrated using examples.