Statistics for qPCR with varying number of samples between groups

Question

[V24]

Hi,

I have done qPCR to check for differential expression of my target genes. I want to compare control (healthy donor stem cells) to different developmental stages (pro B & pro T cells from different healthy donors) and cancer cell lines that match to stem cells and the developmental stages.

Of note is the fact that I have 4 samples of healthy donor stem cells and 6 from pro-B (different donors) and 6 from pro-T (different donors). The samples in all the groups come from completely different donors.

My questions are: 1. Whether I can have different number of samples across different groups? 2. If so, what kind of statistics could I apply to get a significant result? 3. Does it make sense to include the cancer cell lines (though they do match the stages of the healthy donor samples)?

Any help will be appreciated. Many thanks, V

Answer 1

[Devon Ryan]

1. That's not a problem
2. Either a T-test or ANOVA/linear model, depending on whether you want to include the cell lines.
3. I'd be hesitant to do so, mostly because it's not entirely clear what the results are adding.

[V24]

Thanks for your quick response. I'm including cancer cell lines to see if the expression ratio of my targets' (which are actually 2 different variants of a gene) varies under oncogenic stress. Does it still make sense? and, I did two independent experiments with the first 3 sets of samples from the healthy donors (3HSC, 3pro-B, 3pro-T & the cell lines) and the 2nd experiment with the rest 3 of the donor samples (3HSC (2 of them were bad samples), 3pro-B, 3pro-T & the cell lines). My question- can I combine the data together so long as the amplification efficiencies of my targets don't vary between the experiments, right?

[Devon Ryan]

I'm generally hesitant to draw strong conclusions from comparing human ex vivo results to in culturo results.

Anyway, yes you can combine them if it makes biological sense.

[V24]

I'm sorry to put forth all my thoughts as questions. It's just that I have no guidance whatsoever and I do want to make meaningful inferences out of my experiments and not just plot standard errors and asterisks. Thanks for your time.

Answer 2

[V24]

Hi, I have plotted the fold change from different cell lines relative to a calibrator sample. Did a 1-way ANOVA of the data and p value was significant. Does the plot make sense?![enter image description here][1]

[Devon Ryan]

Sure, the plot makes sense.

[V24]

Thanks for your quick response!

Answer 3

[V24]

Hi again, one more clarification. I have data from two independent qPCR experiments; each with 3 different samples from 3 different tissues, so in total, I have 6 different samples from 3 different tissues at the end of the two qPCRs.

My question is, I want to represent the fold change in gene expression across the tissues (one of them being calibrator). How do I combine the data from these two experiments? Finally I want to represent the fold change from all the 6 samples, I followed the delta delta Ct method after checking for efficiency of the target amplifications. Please help.