Butte: Genomics in medicine

Every Thursday there is a seminar on biomedicine informatics on Stanford. I will try and make a serie of summaries of these seminars on my Blog. They might be a bit hard to understand for non-bio-something-informatics but dont hesitate to ask questions!

Yesterday, 6th of October, Butte had a very interesting talk on informatics in diagnosis. He has just startet a project that, as well as I understood is mostly about crosslinking.

One of the problems in recent research is that papers and analysis are allways counting on only one method. Microarrays, SNPs and proteomics are not used togehter when veryfying tests. Very often, the metadata is not used as well. (The data with context-related data, e.g. age, race etc) Of cause, there are some problems in just combining the tests and calculate the probability, but the outcome can easily become more significant. A further discussion about - why - people dont do these things (Cost, culture etc) could be interesting.

Another task he is working with, is combining UMLS into the annotation of microarray. The UMLS is a kind of dictionary, in which you can enter some concepts and get the medical related semantics and Butte wanted to make some crosslink to the GEO-database for getting the concept of the medical terms in GEO-profiles. (Then, unfortunately, I had to go to some other lectures, since I am paying f****** many money for that course)