Can anyone offer any advice on how to decide when to remove a triplicate from the endogenous control or the target genes before analysis? I am interpreting the raw data using excel. It is an mRNA gene expression comparison between two treatment groups (allogeneic) (+/-) the treatment and the control (syngeneic). I used C57 bl/6.
Thanks,
Jim
Hi Jim.
For statistics n=3 is still very low, although common practice in biology. If you remove your triplicate sample because does not look right (outlier) you will have to take into account that these points remain "observations" and you should not just throw them out on a whim, without considering what that would cost to your analysis and most importantly interpretation...specially in a n=3. I would suggest you have good reason(s) to remove your triplicate or any other kind of outlier. In fact, you may have several truly valid reasons to remove data-points...say, outliers caused by measurement errors, incorrectly entered data-points or impossible values in real life...BUT in any case, if you feel (and this is very personal) that your triplicate is an erroneous data point and you can validate this, then you should feel free to remove it.
I hope this helps!
good luck!
Thanks for your answer Luciano. I am developing a criteria for qualitatively removing an outlier by looking at the triplicates of the EC gene. It is tricky, that is why I would like to rely more on stats. For example, if the EC triplicate for a sample is tight, say below a std.dev of 0.8 Ct, but the averaged Ct is more than twice the std.dev for the entire population of the EC gene across all groups, then I guess I would say that something is likely wrong with that EC gene in the sample. Especially using a model with no real genetic variation like C57 bl/6, AND the rest of the target genes comparatively performed with in "normal" values.
So I am still not sure about that 0.8 Ct limit, because what if the triplicates are say 27, 28, 29 with a std.dev of 1? which one do you through out? Of course it makes a difference which you choose, because it affects the triplicate average and the normalized value for gene expression. I am staring to lean towards just using greater than one and not 0.8 Ct
Thanks again for your input,
Jim
- Badges
- Science topic