What percent of reads should align to a reference gene to successfully validate it through RT-PCR?

Question

[Glory Basumata]

I have a list of differentially expressed genes (human) which I would like to validate through wet lab. One of the gene in normal condition do not have any read aligning to the reference genome while the treated sample has quite a lot of reads. I am curious to know is there a specific range or percent of quantification you follow before confirming the finding through RT-PCR.

Attached Image Link: sashimiplot https://ibb.co/y4Zx8Wz

[GenoMax]

Is that result (no reads aligning in a sample) supported by adequate biological replicates? If you have ample replicates (min 3, all showing no alignment in one condition and lots of reads aligning in other) then it would be worth a further check. If you are going on n=1 then you could try it at your risk.

[Glory Basumata]

Hi genomax, Thanks for your response. I should have originally posted with the image. Yes I am working with a replicate (n=3). I have attached a link above with sashimiplot for the same gene. Kindly have a look at it. I appreciate your help.

[Eric Lim]

IMHO, this is more of a biological question than a computational one. Whether it's qval, log2fc, or any other relevant measurement, you'll need a predefined set of thresholds to determine a short list of interested genes. Does the differentially expressed gene make biological sense given the experiment? Do you expect that change given the treatment? If you're measuring gene expression via tarqman qPCR, are you doing it in the same cell or tissue where the sequencing data was generated from? If so, are the 50 or so reads you get in the treated cohorts (out of the unknown number of reads sequenced) enough to be sufficiently detectable?

The computational evidence speeds up the initial discovery process. Translating in-silico to in-vitro/vivo should be biologically motivated. IMHO.

Answer 1

[lshepard]

You should follow the statistical cutoff used for your analysis along with perhaps a fold change cutoff as well (example: p-adj < 0.05 and absolute log2FoldChange > 1). The fact that you have no reads in one of the groups and a lot in the other may likely be due to a biological reason (effect of your treatment), and ideally you should see the same in your RT-qPCR.

[Glory Basumata]

Hi Ishepard, Thank you for providing the details. This gene came from the p-adj <0.05. Although I have to check the log2 fold change value. Please have a look at the attached image link to the post above. Thanks for your help.

[lshepard]

I don't see why you should not validate this gene then.

[Glory Basumata]

Thank you for your suggestion Ishepard.