Gene has covariance close to singular warning, for all genes

[cakrx]

Hi,

First of all, thank you for developing and maintaining this excellent tool.

I’m currently running SPrediXcan.py using outputs from the scPrediXcan pipeline, where I generated my own custom prediction database and covariance matrix. The script runs successfully, but I receive the following warning for all genes:

Gene has covariance close to singular

Could you please advise what might be causing this, or whether it’s something to be concerned about? My understanding is that this may happen when the covariance matrix is near-degenerate, but I’d like to confirm whether this affects the downstream association results, or if there’s a recommended way to regularize to avoid it.

Any guidance or best practices on how to handle this would be greatly appreciated.