Hello,
I have analysed 16S data of cultures that were inoculated from an environmental community (0d=start) in 5 separate cultures and followed over a period of 50 days (17d, 32d, and 50d). Amplicon sequencing was done with two independent replicates from each culture.
The trajectory in the bray-curtis-emperor visualization combines each NGS sample individually along the time scale (17>32>50; 0 is not connected since this is the inoculum).
Is it possible to calculate a 'mean' between two NGS replicates in order to draw the trajecory through the mean and not through each replicate?
I think your quickest solution that retains the most data might be to assign the replicates to multiple trajectories. So, if your transposed metadata looks like this…
Sample
11a
11b
12a
12b
113a
13b
21a
21b
22a
22b
…
Subject
1
1
1
1
1
1
2
2
2
2
…
Timepoint
1
1
2
2
3
3
1
1
2
2
…
Replicate
1
2
1
2
1
2
1
2
1
2
…
(Sorry, it’s easier for visualization to view it this way)
And you’re currently building your trajectory off the subject, I’d create a new column thats a composite of the subject and replicate. So,
Sample
11a
11b
12a
12b
113a
13b
21a
21b
22a
22b
…
Subject x Replicate
1a
1b
1a
1b
1a
1b
2a
2b
2a
2b
…
If you want to be truely random, you could probably randomize the replicates again, but I assume they’ve been assigned the replicate designation at random anyway.
Sure, use qiime feature-table group. You will still need to create a new metadata column that combines time X replicate (or time X site X replicate, depending on whether you want trajectories for each separate culture or not), and group on that column.
