> While I do not question that there were
> differences in gene expression at the end
> of the trial, I have no way of knowing
> whether those differences existed at the
> start of the trial.
We can know by using probability, as this is the most powerful way of
comparing rare and common events.
The paper said "Based on RNA-Seq data, expression levels for 300 genes were
found to be significantly changed in imidacloprid-exposed larvae". What
would the odds be that, when presented with 6 colonies, and randomly
selecting three to be in the test group, that all three would have the same
300 unique altered generic expressions, while none of the other 3 colonies
would? That's not hard, we can view each of the 300 expressions in each
hive as a simple coin flip. So we have to get all the coin flips "heads"
for each of 300 genes in 3 test hives, and we'd have to get all the coin
flips "tails" for each of 300 genes in the 3 control hives.
The probability of flipping a coin and getting:
1 heads flip would be 50/50 or (0.5)
10 heads in a row would be (0.5)^10 or 0.0009765625 (9.7 ^ -4)
300 heads in a row is (0.5)^300 or 4.9 ^ -91
3 * 300 heads in a row is (0.5)^900 or 1.1 ^ -271 that's a very silly small
number, but wait, there's more!
We have to also get all the "tails" right, so we'd have:
6 * 300 correct flips is (0.5)^-1800 that's 1800 places to the right of the
decimal point!
Now my calculator throws up its hands in frustration at such a tiny number,
and calls it zero.
I think it is safe to safe that such tiny probabilities are accurately
described as "essentially zero".
> not supported by the data from the Fraziers, who have
> pointed out that even side-by-side hives can exhibit vastly
> different exposure to pesticides due to their foragers
> foraging in different areas.
Frist, Randy argued that the control hives being 150 meters away from the
test hives was "too far", now he argues that even side-by-side hives would
give us "vastly different" results.
If we accept this new argument on its face, then no pesticide results from
hives in field settings can EVER be valid, as even side-by-side hives would
be "vastly different". But again, we do have statistics to help us, so we'd
have to work out the odds on all the foragers from the control hives
avoiding one or more pesticides, and foragers from test hives bringing the
pesticide(s) home. So, 20K foragers per hive? And 3 hives that showed the
effects, and three that did not, and an average of 6 foraging trips per
forager per day... the odds against are again so small that they also run
beyond the limits of my calculator.
> we have no way of knowing if the tiny bit
> of imidacloprid fed to the test hives had
> anything to do with the observed differences
> in gene expression.
To claim such uncertainty, one would have to explain away why the 3 test
hives not only showed consistent results not found in the controls, but
specifically showed results that were similar to those found when fruit
flies were fed Imidacloprid: "Consistent with those results, RNAi mediated
knockdown of Cyp6g1 renders adult Drosophila more susceptible to
Imidacloprid [28]" and "Knockdown of dHR96 increases tolerance of adult
Drosophila to Imidacloprid exposure [28]".
Overall, the criticisms are outside the bounds of what I would consider
reasonable and substantive, in that they require one to believe in multiple
extremely unlikely events, all happening to this one team of researchers and
to this single group of hives.
***********************************************
The BEE-L mailing list is powered by L-Soft's renowned
LISTSERV(R) list management software. For more information, go to:
http://www.lsoft.com/LISTSERV-powered.html
|
