Generating specifically expressed gene sets for LDSC-CTS

[Jonathan Chung]

Hello,

I would like to apply the cell type enrichment analysis described in the Finucane, 2018 paper on a custom set of gene expression data. And I have a few questions about generating the t-statistics.

• Are the t-statistics generated specific to the dataset you are analyzing? If I wanted to combine the GTEx and our internal data, would I have to re-calculate the t-statistic across all samples? 
• Is there some example code implementing the t-statistic equation shown in the methods section?
• Is there a minimum number of genes needed to accurately calculate partitioned heritability?

Thanks,

-Jonathan