PARC public forum: Exploring Genomic Medicine Using Integrative Biology

I'm right now in the talk called Exploring Genomic Medicine Using Integrative Biology.

Atul Butte, Stanford University School of Medicine

The past 10 years have led to a variety of measurements tools in molecular biology that are nearly-comprehensive in nature. For example, microarrays are just one of at least 30 large-scale measurement or experimental modalities available to investigators in molecular biology. Instead of focusing on the cell, or the genotype, or on any single measurement modality, using integrative biology allows us to think holistically and horizontally. A disease like diabetes can lead to myocardial infarction, nephropathy, and neuropathy; to study diabetes in genomic medicine would require reasoning from a disease to all its various complications to the genome and back. To enable such research, we have been studying the process of intersecting genome-scale data sets in molecular biology, such as those from genetics, genomics, knockout experiments, and many others. I will show how we have built computational tools that reason over these types of data to help enable discoveries in genomic medicine, with specific applications for obesity and diabetes mellitus. Though standards are increasingly being required and used for genome-scale data, representing the experimental context using a structured vocabulary has not yet happened, yet is a crucial step towards automated integrative biology. I will show how the largest unified biomedical vocabulary can now be used to represent microarray sample annotations and show examples of visualization, searching, and analysis using this coding that could not have been done before. I will end with a consideration of ways we can use genome-scale data to provide new ways to classify disease, and show how this broad recasting of disease nosology allows identification of new therapeutic opportunities, and of the specificity, or lack thereof, of disease biomarkers.

Atul is talking about how we can use wafers instead of microarrays to hold millions of probes but each scan takes around 10 terabytes per GIF image. Your entire genome will fit on a CDROM. Genomics can be used for diagnosis of disease. Algorithms are needed to integrate genomes and phenomes. Johannsen's equation in 1908 is phenotype = genes + environment. Instead of doing this one of a time, relate all genes to all aspects of the environment to all phenotypes, and use text mining. Data is already stored in a system called Unified Medical Language System, it already has gene ontology and other biomedical vocabularies, and this semantic network is all free to download for research. You can then extract genes and concepts from the genome-phenome network of nodes and links (like a social network). We can then create a table of human disease genomics collection using hierarchical colection. The same gene can be used to determine whether to use a particular drug or not and what type of disease. We don't have an ontology to store the genomic test results from one hospital or another. Many physicians do not know how to use the genome. Genome-wide measurements are here, managing this data and relating it to medicine is a challenge. Bioinformatics can help in this regard.

This talk reminds me how bioinformatics is a hot area especially in Computer Science of which University of Toronto's Computer Science department is also being involved in this area.