question on longitudinal linear-mixed-effect results summary

Hello Qiime2 team,

I have a question about the summary output of the longitudinal linear-mixed-effects plugin. Below is the command that I entered and attached are the metadata.tsv and the first-distances.qza files. I am using qiime2019.1 (conda environ in Ubuntu 18).

qiime longitudinal linear-mixed-effects --m-metadata-file metadata.tsv --m-metadata-file weighted-firstdist.qza --p-state-column StudyDayN --p-individual-id-column pig --p-formula “Distance ~ StudyDayN*StudyGroup” --o-visualization weighted-lme.qzv --p-metric Distance

I am trying to analyze the first distances with StudyDayN and StudyGroup as fixed effects. The pig column is the id of the sample, StudyDayN is the longitudinal factor, StudyGroup is another fixed factor. After running, the weighted-lme.qzv is successfully created, but the summary section of the results do not look correct. Specifically, the No. Groups and min/max/mean samples per group are not what I expected. Based on the attached metadata.tsv, I expected 10 groups with a min of 5 per group, max of 6 per group and mean of >5 per group. But the results summary has No. Groups as 27 and min/max as 1/2, respectively. Is my formula or individual-id not identified correctly?

[weighted-firstdist.qza|attachment](upload://qRih19MELzwXCI8SSqFWxPDfMDwmetadata.tsv (2.0 KB) .qza) (76.4 KB)
weighted-firstdist.qza (76.4 KB)
Thank you,
w

Hi @wmuraoka,
Good question — in the context of this test “groups” is referring not to your treatment groups but rather to the number of individuals. The min/max per group then is the number of time points.

Since you are running this on first distance results, this all makes sense: not all individuals have all 3 timepoints, so the min should be 1 and max should be 2 (not 3, since first distances is looking at the interval between successive timepoints).

I hope that helps!

Hello! Thank you very much for clarifying. This makes sense now with your explanation.

1 Like