Exercise 6: Fix Experiments with Bad or Missing Data

TODO Convert this to h5p

In my qPCR experiment I want to study gene expression of 6 genes (3 genes of interest and 3 reference genes) in 20 samples (samples of interest and control samples). I want to use 2 technical replicates. One of my genes of interest failed completely and I want to repeat the measurements for this gene in a new run.

Do I need to include IRCs ?
No. We can put the 20 samples of the gene that failed on a single plate so we do not have to include IRCs.
Do I need to include reference genes ?
No. We just repeat all samples for the gene that failed and replace the old data with the new results.

One of the reference genes failed completely.

What should I do ?
Depending on the quality of the two remaining reference genes, you should either do nothing or do the same as in the previous example where one of your genes of interest failed. If the two remaining reference genes are stable you can do the normalization with the two remaining reference genes.

Three samples failed completely.

What’s the first thing I need to do ?
Since they failed completely, they are probably of low quality. Therefore, you have to prepare the samples again, check their quality and then use them for qPCR.
Do I need to include IRCs ?
Yes. If you want to compare these samples with the samples that didn’t fail, you have to perform inter-run calibration.

Three samples failed for one of the genes of interest

What is the first question I need to ask ?
Is the gene expressed in these samples ?
Is it possible the RNA of these three samples was of low quality ?
Not likely, the measurements for the other genes in these samples are ok.

Three samples failed for one of the reference genes

Can I use the measurements of that reference gene in the non-failing samples for normalization ?
No, qbasePLUS requires that you use the same reference genes for all samples so you have to discard all samples for that reference gene.