Monday, August 27, 2007

How not to impress your peers: Crappy gels

I was all excited last week because I thought I found an awesome paper to review for you guys:

Suppression of viral gene expression in bovine leukemia virus-associated B-cell malignancy: interplay of epigenetic modifications leading to chromatin with a repressive histone code.

Sweet, huh? Retroviruses, epigenetics, cancer, WHOOO!!!!

Alas. Theres a really fast way to get people to question your work before they read one word of your paper: publish crappy gels.
I thought I might just be biased, as Im good at running gels, and I like it, but a PI commented "You know, gels are like keeping your shoes tied. If you cant do it, you look sloppy. Its embarrassing."

The authors inability to run nice gels effected their paper. They used P32 to label their nucleotides-- They should have been able to quantify exactly how their transcript levels changed under different epigenetic conditions. Like on my gel I linked to, you should be able to draw a box around each of those bands and say "There is X-fold more Band 4 in lane 2 than there is in Band 4 in lane 3."

But they couldnt because of their sloppy lab work.

But that was the whole point of several figures in the paper-- they tried to show a dose-->response curve with a chemical that effects epigenetics... and the gel just looked like a damn mess. Yeah, you add drug and you got more viral transcripts, but thats a yes/no answer. Running a gel on
P32 labeled nucleotides should and could have given a descriptive answer.

Run a large gel instead of a mini. Skip wells. But for the love of Bowie, dont ruin your experiment, ruin your discussion, and ruin your reputation by publishing a crappy gel. Blech.


Israel Barrantes said...

whoa! this is the kind of nonsense articles that i also like to review on my blog. check those labelled as 'controversial'.
And of course i'll check this paper out later!

Death to fake science!

Ary Shalizi said...

A linguistic quibble... I think you mean "quantitative" and not "descriptive." Saying you have more or less transcript is descriptive, it's just a qualitative description.

ERV said...

Israel-- I mean its *kinda* a good idea, showing how epigenetic control at a cellular level can effect cancer at an organismal level... but they treated the cells before they injected them into the sheep.

And they probably should have studied cows, anyway.

Eh. Just too bad they ruined their data.

Ary-- Well, its more descriptive than Yes/No :) But quantitative is a good word too! P32 can be quantitative!