I have a planned microarray experiment but I'm having some difficulty designing the best possible experiment given the conditions. I have enough for 3 arrays per condition, but six biological sources. The problem is that I can only sample the biological sources once every two months, and there is a chance that the samples will fail.
Originally I was thinking of randomly selecting three animals from which I would draw samples from three times each. That way each array represents one animal but replication of the sample isolation and treatment has been replicated. However, this is a six month process under the best case, if any collections fail this will add a significant amount of time, which is a problem given the year I have to get this work submitted. However, this allows me to control for sample to sample variation within animals and between animals.
On the other hand, I was thinking each array could by a pooled sample from three animals. That way there I still account for sample to sample variation but avoid the lengthy wait associated with the other approach. The drawback here is I don't have much information on the individual animal's responses. While I expect them to be very similar, I'd worry about this becoming an issue later on. An advantage here is that the three arrays become technical replicates of the arrays, which is would result in stronger data.
Of the two options, which is the best approach? Or, if they're both insufficient, what would be the best decision?
What question are you trying to answer?
Also, be aware, if you pool samples from all three animals and stuff them onto three arrays, it's _technical_ replicates, not biological. Any findings from technical replicates are very uncertain, and a lot of people will be very skeptical.
I have two species, I want to compare the responses to the same experimental condition. I have six animals from each species but only 3 arrays per condition.