Guest commentary on chapter 8: Data integration: The next big hope?

Yves Moreau

Katholieke Universiteit Leuven, ESAT/SCD, B-3001 Leuven-Heverlee, Belgium

There is no doubt molecular markers are invaluable for diagnosis, prognosis, and therapy selection and follow-up in numerous pathologies. In fact, many of the classical diagnostics lab measure molecular markers. Examples include oestrogen and progesterone receptors and HER2/ErbB2 in breast cancer or the genetic markers BRCA1 and BRCA2 in familial breast cancer. For the sake of concreteness, I focus on breast cancer as one of the most active areas for complex molecular models, but it applies to many other pathologies. (An example is a statistical or machine learning model combining measurements from multiple molecular markers.)

What has changed in the past decade is the capacity to measure the genome and transcriptome genome-wide and the proteome and metabolome on a large scale. This has led to the expectation that unprecedented insight into the molecular mechanisms of complex pathologies and superior predictive models were just around the corner. The results so far have been humbling at best . . . For example, in breast cancer the main classification of breast tumours remains that based on the oestrogen and progesterone receptor status and the HER2/ErbB2 status. While some more complex models are taking hold, such as the 70-gene signature of van’t Veer and colleagues (Cardoso et al., 2008) or the 21-gene assay of ONCOTYPE-DX ...

Get Bioinformatics and Biomarker Discovery: "Omic" Data Analysis for Personalized Medicine now with O’Reilly online learning.

O’Reilly members experience live online training, plus books, videos, and digital content from 200+ publishers.