Identifying proteins with related structures and functions presents another major problem with implications for function and evolution. Recent progress enumerating the protein kinases in the human genome provides a dramatic example of the challenges. Protein kinases are ubiquitous regulatory enzymes. Nearly 25% of drug development programs worldwide currently target kinases. For example, the recent breakthrough in treating chronic myelogenous leukemia was achieved by targeting a specific kinase that is activated inappropriately in the cancerous blood cells. Three months ago, an authoritative winnowing of the human genome sequence placed the number of protein kinases at slightly over 400. Last month, in an article in Science, a more sophisticated algorithm was used to place the number at 518, an increase of over 100 known human protein kinase genes in just four months! The latest report represented a tour de force not only for the computational advances involved, but also because a Bay Area biotechnology company verified experimentally over 80% of the kinase genes that were identified computationally. Such a large-scale experimental analysis is practical for only a small number of gene families, placing an increased burden on computational methods to identify the vast majority of protein families.