tips to study rare variants

[BookwormBro]

I'm up to my eyeballs in this project, trying to find those rare variants linked to this phenotype I'm interested in.

now, I'm sitting on about 200 cases and 50,000 controls with exome sequencing data. I'm thinking of using the burden test to see if rare variants are popping up more in the cases compared to the controls.

But with this small case number, I'm just not sure if I have enough power to actually spot anything.

So, how do I figure out the power I have?
And is there a better study design I should wrap my head around to find these rare variants given the sample sizes I'm working with?
all your tips and advice are welcome, TIA

 

[|Wind_Whisperer00|]

Yeah bro 200 cases for rare variants is an uphill battle I had a similar numbers game once. For power there are tools in R you can use and maybe also look into how you are filtering or grouping variants to give your analysis a bit more punch.

 

[SnipSniper]

I think the burden test is a great idea! But honestly, 200 cases might be cutting it a bit close when it comes to spotting those rare variants. You might want to do a quick power calculation with tools like the Genetic Power Calculator or RVTests. Have you looked into SKAT-O as well? It's actually pretty good for those smaller sample sizes with rare variants. By the way, what kind of effect size are you hoping to detect?

 

[emma w.]

I'm up to my eyeballs in this project, trying to find those rare variants linked to this phenotype I'm interested in.

now, I'm sitting on about 200 cases and 50,000 controls with exome sequencing data. I'm thinking of using the burden test to see if rare variants are popping up more in the cases compared to the controls.

But with this small case number, I'm just not sure if I have enough power to actually spot anything.

So, how do I figure out the power I have?
And is there a better study design I should wrap my head around to find these rare variants given the sample sizes I'm working with?
all your tips and advice are welcome, TIA

You should try SKAT-O! It's great at handling sparse signals better than a straightforward burden test. Also, grouping variants by gene or function might be the better route for you as well.

 

[X-Man]

@BookwormBro I'd second what @emma w. mentioned, SKAT-O can sometimes squeeze more out of smaller case numbers than a straight burden test. With only 200 cases, grouping variants functionally could give you a clearer shot at picking up signals.