Part III – Computational Platforms
Addressing important clinical questions in cancer research will benefit from expanding computational biology. There is a great need large amount of information to improve prevention, early diagnosis, cancer classification, prognostics and treatment planning, and to dis- cover useful patterns.
requires integrated and systematic approach. We still lack under- standing, and we are ramping up technologies to produce vast amounts of genomic and proteomic data. This provides both the opportunity and a challenge. No single database or algorithm will be successful at solving complex analytical problems. Thus, we need to integrate different tools and approaches, multiple single data type repositories, and repositories comprising diverse data types.
Knowledge management is concerned with the representa- tion, organization, acquisition, creation, use and evolution of knowl- edge in its many forms. Effectively managing biological knowledge requires efficient representation schemas, flexible and scalable retrieval algorithms, robust and accurate analysis approaches and reasoning systems. We will discuss examples of how certain repre- sentation schemes support efficient retrieval and analysis, how the annotation and system integration can be supported using shareable and reusable ontologies, and how to manage tacit human knowledge.
Data from high-throughput studies of gene and protein expres- sion profiles, protein-protein interactions, single nucleotide polymor-
each of the areas separately. The challenge is to use novel approaches that efficiently and effectively integrate and subsequently mine, Bill Wong and Igor Jurisica
statistical, machine learning and data mining approaches have analyzed to support systematic knowledge management and mining of the
phism, and mutant phenotypes are rapidly accumulating. Diverse
Understanding normal and disease states of any organism
86
visualize and interpret these various levels of information in a sys- tematic and integrated fashion. Such strategies are necessary to model the biological questions posed by complex phenotypes, typi- cally found in human disease such as cancer. Integration of data from multiple high-throughput methods is a critical component of approaches to understanding the molecular basis of normal organism function and disease.
Cancer Informatics in the Post Genomic Era