When reading in weights from a .db it can be inconvenient to have rsid hard coded as the SNP identifier. Especially since there is another column, varID, which could be used. While working with predictdb mashr gtex v8 eQTL and sQTL models, I've used that identifier instead. Some of the SNPs in these models, if rsid is missing for whatever reason, use varID as a backup in the rsid row anyway as well.
https://github.com/xinhe-lab/ctwas/blob/47c0e3b114799af5cf2ef86de2569b58f78e0751/R/ctwas_read_data.R#L434
When reading in weights from a
.dbit can be inconvenient to haversidhard coded as the SNP identifier. Especially since there is another column,varID, which could be used. While working with predictdb mashr gtex v8 eQTL and sQTL models, I've used that identifier instead. Some of the SNPs in these models, ifrsidis missing for whatever reason, usevarIDas a backup in thersidrow anyway as well.Here's an example of a
db_idparameter added to allow the user to specify the SNP identifier from a.dbweight source. https://github.com/shugamoe/ctwas/commit/bd42fc0924bf6b944f4338c2ec5abf108e80b62a