Dear all,
I have an RNA seq data and I have many genes whose expression are altered by my treatment. However, The up-regulation pattern is generally about %50 (which makes Log2foldChange value 1.
I am quite sure that, these genes are realy associated with the downstream and upstream pathways of my treatment. However the changes are not extreme ( I am also sure about that my treatment should be resulted in a slightly change on transcriptom., my p values are below 0,05)
In that point for analysis can I use a threshold lower than 1,5? Because 1.5 makes 3 fold change which is not possible with my treatment.
For example can I use a threshold for the genes which are over-expressed 20%.
Thank you for your help.
Yagiz
You can, but then you have to ask yourself, what would be the biological significant of such a change. For example, if you 1.5-fold increase in gene expression would you expect to see a big change in cellular response if only have half again as much protein - recognizing, too, the variance in expression. For example, you may even have individuals in your control group where the gene(s) in question are also 1.5 x above mean. Hopefully, once you start considering variation, you will realize that log2fold change is an entirely arbitrary value.
Hi colleague
find the following URL may help you:
https://www.ncbi.nlm.nih.gov/pmc/articles/PMC4934518/
Regards
Hello
Yes you can, in case of lncRNA or miRNAs, they have a great impact other genes if their fold change is high. They are regulators of other genes and with less than 2 fold change, they may probably have a big effect on other genes.
As Mark Farman mentioned, you need to think about it thoroughly.
Thank you all for your responses and detailed answers. I will use a lower threshold than 1.5.
Of course, you can, but is dependent on your scientific answer. For example, RNA interference, Transcriptional Factor, has a little change, and some cases undetectable by statistical methods but has greater biological importance. The log2FC or adjusted p-value threshold is arbitrary. Why people select 2, 3, or 5 times to differential-difference? I know why we use some threshold but dependent on the subject, but if we observe a rare event, we need to increase the sample size. In my transcriptome analysis, I prefer using GSEA (Gene Set Enrichment Analysis) for functional analysis, mainly because the tool is independent of the threshold.
Definitely you can. Its depends on the objective of your study design, and the genes that you want to emphasis.
- Badges
- Science method