what are the reasons to find so few Differentially expressed genes (DEGs)?

Question

[sunnykevin97]

Hi

I had sample size of 32 (8 X 4 = 32), 3 treatment groups and 1 control. I used edgeR pipeline, to find the differential expressed genes (DEG's). At the end of the analysis I find only 8 DEG's with greater than 1.5 fold-change ratio.

I'd like to know the reason(s), why the RNA-seq experiment doesn't capture the signal?

I'd like to state the reason in my research article, help is needed.

I tried with different pipelines (DESEQ2, voom), the results are same, irrespective of the pipeline used.

Thanks
Kevin

[Ram]

Why do you expect to see more than 8 DE genes? Do you have reason to believe other genes should be DE?

[sunnykevin97]

If the gene is said to be differential expressed it means alt-least it should have greater than 1.5 fold-change ratio. I find only 8 genes with 1.5 FC ratio.

Suggestions.

[sunnykevin97]

From the physiological experiment we able to find good results in understanding the animal behavior during the stress

treatment. But we unable to find the DEG's related to stress behavior (serotonin and dopamine).

[sunnykevin97]

May be its because of the "phenotypic plasticity" which triggers more than one phenotype while exposing to the different treatment (environmental conditions).

[h.mon]

With the amount of information you provided (and assuming your bioinformatic analysis is without errors), there are only two possibilities:

1) Your data got messed up somehow (sample swaps, poor quality RNA, and so on).

Did you perform the usual quality checks for your samples?

2) the analysis reached the correct conclusions, but it is not what you expected.

Biologically, do you expect large differences in your treatments? Why? Maybe the effects are indeed small, and the results are reliable.

A log2FC of 1.5 is arbitrary - as any other value, really, but often times is useful to focus on the "very" differentially expressed genes. What happens if you lower your threshold to 1? Or don't set a lof2FC threshold at all?

[swbarnes2]

The OP might literally mean a fold change (not log2) of 1.5.

Answer 1

[seidel]

There's no way to really answer this without knowing more about what the experiment is, but it looks like you have 4 conditions, and 8 replicates. Do you have a lot of variation between your replicates? Are your replicates more correlated with each other, than with replicates from the other conditions? If you have a lot of variation between replicates, you will have a hard time finding DE genes. You can use a correlation table to examine this. On the other hand, perhaps your experimental conditions were not sufficient to cause any genes to be differentially expressed. If you have no spread in your data between conditions - perhaps that's the result of your experiment. Imagine blowing on tubes of cells. That might count as an experimental condition, but hopefully not one that would cause a biological repsonse. Do you have a control gene that you know responds to your conditions? Did it respond? Do you have a positive control condition that you know invokes a given response? Did it respond?

Answer 2

[Mensur Dlakic]

I'd like to know the reason(s), why the RNA-seq experiment doesn't capture the signal ??

Assuming everything was done correctly during data collection and processing, the answer would have to be in the domain of biology. Sometimes there is no signal to be captured.