Hello,
I would like to perform sPLS analysis using the N-integration of microbiome data and several clinical continuous variables. Can I use the relative abundance of microbiome data (Total Sum Scaling - TSS) as input for X without applying a CLR transformation? Or, do I need to perform the CLR transformation on the data before running the function? I have some zeros in my TSS data.
Thank you!
Hi @Tonima ,
Yes you can do this, then check that your TSS data don’t have a zero variance (otherwise filter them out). Zeroes are no problem if it’s not across all samples. But generally we don’t really advise using TSS, but that’s up to you.
Kim-Anh
Hello,
Thank you so much for replying !
I have another question. My study is longitudinal. 8 subjects have 5 samples from 5 timepoints (8*5=Total 40 samples). For tune() and perf() function, is “Mfold” going to be appropriate choice? if so, should I choose (40 samples/5) 8 fold validation, or (8 subjects/4)= 2 fold validation.
Thank you!