Much research in statistics is focused on very applied problems in computational biology. You are right that statisticians do not perform the experiments; it's unrealistic to expect them to have the lab experience necessary. The mathematical statisticians are working on problems such as multiple testing (in many studies there are hundreds of thousands of hypotheses being tested) and inferences of very high dimensional data. These are relevant topics and the work is motivated by the recent shift in the types of data we collect in experiments.
Some interesting subsets of computer science (machine learning and AI) are now focused on statistical models. You have pointed out that much statistical work isn't developed by statisticians. That doesn't mean the field is dead -- it is thriving!
Some Bayesians who rely on MCMC and may not care as much, but generally it is still important to find models with closed form likelihoods and optimization updates. Applied statisticians work hard to keep things in closed form precisely because it matters in practice (but not in theory).
> This GCV is used in Q-value routine of 2003 to determine false discovery rates by smoothing over the P-values of thousands of genes
Are you referring to this paper ?
http://www.genomine.org/papers/Storey_Tibs_PNAS_2003.pdf
The authors didn't mention any numerical difficulties. Maybe it would have been worth contacting one of them for help--assuming you're not one of them.
I saw no mention of the EPA in the article.
However, were it not a troll, the "Try again." comment would bother me. Someone may have little to contribute to the thread except for a link to a journal article. Let them post it! It still gives information to a non-negligible subset of the slashdot readership.
http://autos.msn.com/advice/CRArt.aspx?contentid=4023544
The American cars are apparently more reliable than you remember.
"The one charm of marriage is that it makes a life of deception a neccessity." - Oscar Wilde