Rarefy/Sampling Depth Question with various days

I am gearing up to run some core diversity analysis on a set of shotgun data in which I have utilized Metaxa2 to develop and abundance table. My question pertains to how to set sampling depth. I am new to QIIME2 but have used QIIME1 in the past, several times and always struggle with this. Many of the posts suggest utilizing something like the median frequency/sample. My biggest question is in my dataset I am comparing the microbiome at days 1, 3, 28, and 180. As you can imagine the median sequences/samples is MUCH different between these time points. Are there any suggestions for how to go about this type of issue?

Any guidance is appreciated!
Thank you!

Hi @hcunnin6 - unfortunately there are no hard and fast rules for selecting an even sampling depth. I would highly recommend checking out feature-table summarize - if you provide your metadata file, the second tab will display a really cool interactive rarefaction visualization (e.g. moving pictures tutorial). Here you can interactively see the impact of a particular sampling depth on your samples, and the metadata values associated with your samples. In the example from MP I linked to, set the drop-down to "BodySite" and then set the sampling depth to 5200:

If you choose 5200 as a sampling depth on these data, this plot means that you will lose all of your "left palm" samples - this may or may not impact the outcomes of your study, and as such, is highly dependent on your goals for your investigation (this is why there are no hard and fast rules, among other reasons).

Hope that helps, and let us know what you do! If you want to share a summary QZV here we would be happy to take a look at it with you! :t_rex:

Thank you for that advice. I am unable to run Qiime2 right not but have Qiime1 on a virtual box running on my computer. My biggest challenge is when I rarefy around median then I lose all data for one of my sampling days. When I decrease the rarefaction depth then I get a ‘Common returned exit status: 137 stderr killed’ and it doesn’t complete the run. I am trying to convert my data to proportions and see if that helps. I realize I won’t be able to get group significance anymore but perhaps I can still get alpha and beta diversity.

I plan to get QIIME 2 up and running on our system this summer and will try some other things out then. But for now…QIIME1 will have to do.

Thank you!

QIIME2 can also be installed on a virtual machine so might be accessible to you sooner if, e.g., admin privileges don’t constrain you from using QIIME2.

In the mean time, please post QIIME1 questions to the QIIME1 forum — we are not able to provide support for QIIME1 questions on this forum, e.g, this error you are having:


Sounds good! I’ll work on getting QIIME 2 up and doing.

Thank you!

1 Like

This topic was automatically closed 31 days after the last reply. New replies are no longer allowed.