I'm using SAIGE to perform my GWAS.
Let me explain my situation briefly, I have multiple traits for the same individual sets (N<1000).
Those traits are related to image features (Nfeatures<95).
They are something like fractions ranging from 0 to 1, and I already filtered some features with too many zero-counts.
In this situation, I could get too high Varratio_NULL like 29 from only one feature...
All the other features look fine, they all have VarRatio_NULL<1.
In my understanding, SAIGE calculates taus until the second value of variance component estimates becomes 0, and then VarRatio_NULL is determined by 1/tau[1].
Here is my tau that SAIGE estimated from the strange feature...
This is tau from one of the other features which looks fine.
Could you figure it out?
The feature itself looks okay.
Hi,
I'm using SAIGE to perform my GWAS. Let me explain my situation briefly, I have multiple traits for the same individual sets (N<1000). Those traits are related to image features (Nfeatures<95). They are something like fractions ranging from 0 to 1, and I already filtered some features with too many zero-counts.
In this situation, I could get too high Varratio_NULL like 29 from only one feature... All the other features look fine, they all have VarRatio_NULL<1.
In my understanding, SAIGE calculates taus until the second value of variance component estimates becomes 0, and then VarRatio_NULL is determined by 1/tau[1]. Here is my tau that SAIGE estimated from the strange feature...![image](https://github.com/saigegit/SAIGE/assets/41933720/63985c55-18a6-4b18-8a4c-a500da4fd8ec)
This is tau from one of the other features which looks fine.![image](https://github.com/saigegit/SAIGE/assets/41933720/441159f1-affb-4583-9983-b8623d155d2c)
Could you figure it out? The feature itself looks okay.
Thanks, E