Using CONQUR-corrected matrices in MaAsLin 3

Downstream analysis and statistics MaAsLin
1

Hi BioBakery Team,

We pre-corrected our raw microbiome count matrices for batch effects using CONQUR. We are now analyzing associations with clinical outcomes using MaAsLin 3.

To validate our pipeline, should we omit Batch from the MaAsLin 3 formula given the prior CONQUR correction, and are normalization = 'TSS' and transform = 'LOG' still the appropriate settings for this adjusted input?

Thanks for your support!

2

Hi,

The outputs you get from CONQUR should be similar to what we usually get with MMUPHin. With MMUPHin (and therefore presumably with CONQUR), we usually use TSS, LOG, and we still include a random effect per-batch in the regression in case there’s dependence leakiness.

Will