Back in 1992 I was in a conference where someone famously powerful was proposing to sequence the genome of the model organism with which I used to study my favorite questions in research. Sitting in the dark hall I quickly made a back-of-the-envelope calculation: how long would it take to discover the function of an unknown gene, known only by its sequence, given the rate of gene function discovery for the first organism's DNA ever sequenced completely (the lambda virus) in 1979. The answer was several hundred centuries. So at the end of the talk, I announced the result of my calculation, and asked why not spend the money, millions of dollars, on simple investigator-initiated research that asks straight questions and gets straight answers. There was silence in the hall, and I felt smug and smart. Until the last talk in the conference, which announced the discovery of microarrays, for parallel measurement of RNA transcripts made from hundreds (at that time) of genes. Quite uncomfortably, I realized my mistake--the vice of linear extrapolation. That changed my mind.
So today I read and discussed a paper published this last January, which reported the successful identification of a gene that appears to harbor the causative mutations for Miller Syndrome (Google it, if you don't know it), a particularly debilitating rare and inherited disease that severely affects the lives of some children. The methods that the researchers used, based on sequencing nearly the entire protein-coding gene set of the patients, were unimaginable back in 1992--it would have been considered a science fiction dream in popular forums and would have been derided as madness in scientific company. But it is a reality today. What surprise will the next month bring?
[Update, April 16: Here is a link to an editorial on this paper published in today's Nature Genetics: http://www.nature.com/nrg/journal/v11/n5/full/nrg2783.html]