[Arabidopsis] technical replicates vs biological replicates

hnbjed via arab-gen%40net.bio.net (by hnbjed from yahoo.com)
Mon Feb 9 01:17:53 EST 2009


Hi,
=A0
I often saw many research papers, including those in Plant Cell, Plant Phys=
iol and Plant J, published results with only 2-3 technical replicates, NO b=
iological replicates at all. Obviously, one would get very small standard e=
rror with only tech reps. Many, not just some, reseachers even did not perf=
orm strict tech rep at all. For example, for quantitative RT-PCR, they just=
 extracted RNA from one boil rep (they only prepared tissue once) only, and=
 then ran three parallel rxns, and then say =93data from three replicates..=
=94. No body knows what are values of their data. I will certainly doubt th=
is and challenge them if I am one of the reviewers. After statistical analy=
sis, data from only tech reps are usually significant, cause the SE is smal=
l. But if running bio reps, you may find that your data is no longer useful=
, cause SE is large and thus p>0.05, then you may not dare to say "under xx=
x condition, one gene/mutant is significantly different from the
 other/wt...". In our lab, we have been performing 3 independent biol reps =
(3x3=3D9 measurements), and therefore, I found out that the SE is large eve=
n if I tried my best to decrease the variance (judged by eyes) between reps=
...... Our way, of course, put our papers under a rather disadvantageous conditi=
on. Everyone want to publish papers in a high-impact journal, but we should=
 play fairly.
=A0
Another thing confusing me is why data with T1 generation (overexpression, =
RNAi, and promoter-GUS fusion) was accepted by a recent paper on cell wall =
in Plant Cell? I checked a few earlier papers from the same group in a Geor=
gia university and was astonished to find out they have been using data fro=
m only T1 transgenic plants for their papers in the past years.=A0Is this a=
cceptable? In our group, we at least used T2 generation, usually homozygous=
 T3 generation.
=A0
I=A0am afraid to irritate some researchers, but want to=A0expose some pract=
ices that I do not agree with to every researcher.=A0 I think for biologica=
l experiment, a strict boil reps (at least 2) should be followed.=20
=A0

=0A=0A=0A      


More information about the Arab-gen mailing list