Hi @himel.mallick @fbeghini- The outputs from metaphlan are not normalised to the library size and they are in the form of relative abundance out of 100%. Considering these two facts, can we directly use the merged outputfile from metaphlan directly into MaAsLin2 with the default parameters? or, should we change some options?
Hi @DEEPCHANDA7 - here is a tutorial which I hope will be helpful for some of the analysis-specific questions you have going forward. To answer your specific question, the default model should be appropriate but you may want to subset the data to species-level features only. Thanks!
Yes sir, I have run metaphlan output only with the species level features. I was asking as we are using the same default parameters for both humann output and metaphlan output whereas relative abundance in metaphlan is out of 100.00 and in humann out of 1.00.
@DEEPCHANDA7 The proportional nature of the data mostly matters for transformations like AST (arcsine square root) which is only defined for values between 0 and 1. Since the default model uses log transformation, you should be fine with the percentage data. Even if you convert the percentages to proportions, the results should be similar.