How to filter the relative abundance table with > 0.1% in order to fill in the NMIT?

[tingting_huang]

Hi, I'm currently running NMIT in qiime2. As recommended, "genus-level taxa that exhibit a relative abundance > 0.1% in more than 15% of the total samples" should be fulfilled. I have tried the plugin function "qiime feature-table filter-sample --p-min-frequency" with the relative abundance table which was obviously failed due to the requirement of "a valid integer" means that frequency table can be applied only.
May I ask how do you make "genus-level taxa that exhibit a relative abundance > 0.1%" in qiime2 for running NMIT consequently?
Thank you in advance!

[Nicholas_Bokulich]

@tingting_huang

As recommended, “genus-level taxa that exhibit a relative abundance > 0.1% in more than 15% of the total samples” should be fulfilled

That is not a recommendation, just a statement of what is being performed in that tutorial example.

@tingting_huang

I have tried the plugin function “qiime feature-table filter-sample --p-min-frequency” with the relative abundance table which was obviously failed due to the requirement of “a valid integer” means that frequency table can be applied only.

Indeed. Instead, use that same command on your FeatureTable[Frequency] table to filter by occurrence in a percentage of samples; and use feature-table filter-features to filter by total frequency (you can use feature-table summarize to figure out what 15% of total # of sequences would be).

Then use feature-table relative-frequency to convert to relative frequency.

If you have any doubts, you can always see how the artifacts in the tutorial (or indeed any QIIME2 artifacts) were created by looking at their provenance. Download that ecam_table_taxa.qza, open in https://view.qiime2.org/, and click on the provenance tab. Scroll to the bottom of the provenance plot, and we can see the exact commands and parameters used to make this table:

1. feature-table filter-samples
2. taxa collapse
3. feature-table filter-features
4. feature-table relative-abundance

I hope that helps!

[tingting_huang]

Hi Nicholas, that's really help! Thank you!
For the statement "> 0.1% in more than 15% of the total samples", I understand it should be case specific. May I ask if there's any suggestion about "threshold" consideration in NMIT?

[Yilong_Zhang]

Hi Tingting:

The purpose of the cut-off is to perform the test based on representative taxa. It would be better to try different set-up combinations and identify the one proper for your data.

Best,

Yilong

[tingting_huang]

Hi Yilong, I totally agree with you. I'm running NMIT with the same cut-off as mentioned in the tutorial and will try different set-up later on. Thank you!