Open WenChangYu opened 3 days ago
Hi, You could either use the gene expression imputation models tainted by others. Most people share their trained models, essentially eQTL weights. You can also train a set of models using your own gene expression data and WGS genotype data.
For plant population, you might have the same relatedness as the mouse species. Please see this paper, https://www.biorxiv.org/content/10.1101/2022.06.03.494719v5.full . You might need to use a Linear Mixed Model to train eQTL weights. You would just fit a Linear Mixed Model by GEMMA, https://github.com/genetics-statistics/GEMMA , taking gene expression as response variable, cis-SNP genotype data as predictors, considering additional available covariates. Get the coefficients that are the eQTL weights. These Weights are also suitable for SR-TWAS.
Thanks for your kindly reply.
Dear developer,
We would like to perform TWAS analysis on a plant population. Our data consists of phenotypes, whole-genome gene expression levels, and genotypes for each sample.
We do not have a reference dataset.
Would our data be suitable for analysis with SR-TWAS?
many thanks,
Chunyang