Tuesday, February 25, 2014

Predicting the Future: The Upcoming Stephanie Events

You probably recall this entry on The Steamrollers, technologies that are improving much faster than Moore's law if not more. It turns out that at the Paris Machine Learning Meetup #5, Jean-Philippe Vert [1] provided some insight as to what happened in 2007-2008 on this curve:

Namely, the fact that the genomic pipelines went parallel. The curve has been going down ever since but one wonders if we will see another phase transition.

The answer is yes. 

Why ? from [3]
However, nearly all of these new techniques concomitantly decrease genome quality, primarily due to the inability of their relatively short read lengths to bridge certain genomic regions, e.g., those containing repeats. Fragmentation of predicted open reading frames (ORFs) is one possible consequence of this decreased quality.

it is one thing to go fast it is another to produce good results. Both of these issues of speed and accuracy due to short reads may well answered with nanopore technology and specifically the fact that entities like Quantum Biosystems Provides Raw Data Access to New Sequencing Technology and I am hearing through the grapevine that their data release program is successful. What are the consequences of the current capabilities ? Eric Schadt calls it the The Stephanie Event (see also [4]). How much faster are we going to have those events in the future if we have very accurate long read lengths and attendant algorithms [2, 5] ? I am betting more.

No comments: