Differential gene expression

Question

[4732b159]

Hi all, I am just a bit concerned about my data and trying to understand what that means, could I please get insights from experienced fellows? and thanks in advance! In a differential gene expression analysis can it be possible that the number of up regulated genes is zero while there are over 200 genes down regulated in a diseased tissue? I have checked my code and model multiple times it looks like all is good. Please let me know what you suggest? all the best!

[ATpoint]

Sure can, but one would need some plots, such as plotMA to have an idea.

[4732b159]

thanks for your response :) could you elaborate how to interpret that then?

[ATpoint]

you could add them first here

[4732b159]

So,[ plotMA(res, ylim=c(-2,2)] it looks like that

Answer 1

[OmnibusX]

From my experience, this usually happens when I compare data from different studies, technology, or alignment tools. In these cases, the number of genes in the 2 groups is not equal, for example, in group 1, the technology can capture only 10k genes, while in group 2 the technology can capture 20k genes, during differential expression analysis, genes that are present in group 2 but not in group 1 may appear significantly downregulated. This could potentially skew your results toward downregulation. If you could provide more details about your samples, it would help in diagnosing the origin of this issue more precisely.

[4732b159]

Hi Thanks for your reply :), really appreciated! The data that I am analyzing is from the same experiment of bulk RNA and I have two samples; disease and control. I am particularly interested in what the difference are between the same part of the tissue of disease vs control. I hope is it clearer now. Would you expect in this case? Thanks!

[OmnibusX]

In case you only have two samples, which method did you use for DEG analysis?

[4732b159]

I used DESq2 and designed it as below:

count_data <- as.matrix(count_data)

dds _t<- DESeqDataSetFromMatrix(countData = round(count_data), colData = col_data, design= ~group)

dds <- DESeq(dds_t)

res <- results(dds, contrast=c("group","Disease","Control"))

extra info: The group column of col_data is a factor and consist of Disease or Control.

{{{all(colnames(count_data) %in% rownames(col_data) ==> TRUE }}}

Thanks so much!

[OmnibusX]

Sorry that I might not help in this case, as for the statistical test, I think we need more samples in each group (at least three) for a significant result. Therefore, I can not conclude anything yet.

[4732b159]

Yes I do have more than three samples in each group: disease and control samples, more than 14 samples actually!

[OmnibusX]

I saw your MA plot and code. Everything looks fine. This outcome might be because the disease state is characterized by a general suppression of gene activity or specific pathways, rather than activating different pathways.