Bootstrapped Genome Project (iii) – on Mindless Triplicate Experiments

I get very angry, when someone tells me that all experiments need to be done in triplicate before we can submit a paper. It rubs even more, when he adds - ‘surely you may get by with a lesser journal without triplicate, but to publish in Science or Nature, everything needs to be done three times.’ It saddens me to see plenty of mindless triplicate experimentation going on everywhere, when the money could be spent more cleverly.

In science, there is nothing magical about the number three. If I want to determine the average height of Greeks, nobody would accept my paper with data from only three Greek guys or even five. On the other hand, first human genome paper was accepted with genome sequence of only one earthling. A scientific paper should be judged by how much novel information it presents.

RNAseq is very powerful, and one can extract lot more information by designing the experiment a bit more intelligently. At an early stage of our project, we were discussing a tissue differentiation measurement with a bioinformatician, and he suggested that we do RNAseq of tissue A and tissue B three times. Please note that this was a novel organism with no genome or genes known. So, even publishing the gene sequences would advance the field significantly.

Instead of choosing mindless triplicates, we redesigned the RNA seq experiment in following way. Triplicate measurement was done, but in three closely related organisms with similar tissue differentiation. That way, we would get -

(a) Gene sequences in three related organisms,

(b) Three sets of differential expression data for the same phenotype,

(c) Ability to place the observed phenotype in an evolutionary context.

Think about it yourself - which experiment provides more useful large-scale data? One that gives you genes and gene expression in heart and muscle of three human beings, or one that gives you genes and gene expression in heart and muscle of human, chimp and gorilla measured only once? For argument’s sake, assume that no human, chimp or gorilla genome or genes is known yet.

Privately we discussed the idea with few friends. They all agreed that the second way was far more informative than the first, but also added that doing triplicate as in first alternative was ‘safe to pass reviewers’ of high- visibility journals.

What do you think?