Alpha Diversity q2-longitudinal question

General Discussion
1

Hi QIIME 2 community,

I am trying to statistically analyze my alpha diversity data (including Shannon's Index, Chao1 Index, and Observed Features) from a study I detailed here. In a nutshell, it is a repeated measures design with 12 subjects. We collected feces and swabs from all subjects across 4 timepoints and intestinal samples from half of the subjects (that were cannulated) across the same 4 timepoints.

I'm trying to compare each of the metrics between sample types and across time.

I thought to use linear mixed models for this. However, I checked the normality of the data (from a compiled metadata file obtained from metadata tabulate) using Shapiro Wilk test in SAS. Observed features was normally distributed but the other two metrics were not based on this test. My questions are as follows.

  1. Do I need to transform the non-normal data prior to analysis? If so, what are your recommendations on which method to do that?
    2.Does the q2-longitudinal plugin perform any transformation?

Any guidance is appreciated!

Thank you in advance!

1 Like
2

Hi Mahasti,

Great question. I can't answer all of it (I'm still very new to QIIME 2), but I can weigh in on some of it.

My understanding of linear mixed effects models is that the dependent variable that you pass to the model does not have to be normally distributed. Rather, what is important is the normality of the residuals of the model. There's a really good tutorial that I read a while back here that might be helpful in understanding the assumptions of linear mixed effects models.

Again, I don't think this completely answers your question (e.g., I'm not sure how to check model residuals in QIIME 2), but hopefully it gives you something else to look into.

Hope this helps some!

2 Likes
3

Hi @bioinfx,

Thank you for your quick response and for the link, I'll give it a read to hopefully get more clarity on this!