Hello
I am doing a study about transcriptomic profile of peripheral blood mononuclear cells in patients with liver cirrhosis. I want to see the differently expressed genes in patients compared with healthy controls.
I have a problem with choosing the samples. Indeed, I dont know how many patients and controls do we need to recruit for RNA-sequencing study normally? I have checked through some papers: some include around 20 for both groups (patients and healthy controls), but some use up to 100 individuals.
Because I just have a limited fund, maximum for RNA-sequencing of 20 individuals. So, are 20 individuals sufficient for the study? And, importantly, what are the good points to give reviewers in case they ask about the number of individuals recruited when the disease (liver cirrhosis is complicated with variability of causes, ages, sex and disease developments)?
I haven't mentioned that I also have validation experiment to confirm results obtained from RNA-sequencing using qRT-PCR.
Any advice would be appreciated.
Thank you and regards
Trang Nguyen
Hi Huyen
I believe limited funding is definitely an issue in a lot of places. However, for your study what you could possibly do is move ahead with your 20 samples. Equal number of patient and controls is recommended. Make sure your inclusion parameters are very specific and cover all aspects of the disease. Keeping a doctor in loop would benefit. You can only study one variable at a time in order to get some result. For example
Article Hepatic transcriptome signatures in patients with varying de...
Once you find some concrete signatures you can then go ahead to validate them on 100 more samples using qRT-PCR. For that you can collaborate with some other institute or put it up for separate grant. You dont need to go for transcriptome for all 100 everytime but you do need to validate on a statistically relevant number of patients to get a good result.
You question strongly depend on how your cohort is homogenous (especially test cohort): Patient effect will be the main source of variation, so strict control of experimental bias (biopsy, RNA extraction, ...) will allow you to get more sensitivity for differential analysis. Keeping this in mind, I think starting with 20 samples in each group sound reasonable.
Bests
Dear Guillaume Meurice and Garima Ayachit ,
Thank you very much for your answer. In fact, our sample is peripheral blood lymphocytes, and we already tried to exclude variability such as age, sex and same etiology.
However, my concern again is that how could you choose number 20 or 30? You have to make some calculation, right? Or you may have to cite from some scientific source to say that it is statistically significant. Otherwise, it's not convinced for reviewers when just to pick 20 or 30 samples. In this case, I'm looking an explanation for why you choose that number?
Best regards,
Huyen
Usually a pilot study takes n=20 for study.
Check this link
https://www.researchgate.net/post/How_many_samples_are_needed_for_expression_analyses
As i said earlier you will have to validate it on more samples
- Badges
- Science topic
- Similar topics
- Papers