Minimum count fraction filtering

[Neus]

Dear all,

I have some doubts about how the option --min_count_fraction (from filter_otus_from_otu_table.py) works. 

Just as an example: Imagine that you have one OTU which is the only one present in a sample. Would it be removed if you apply a filtering of 1% and its fraction regarding the whole dataset is 0.6%?

Is there a script to filter OTUs taking into account the minimum count fraction of each sample but using the .biom table of the whole dataset? I mean, without need for filtering samples from the out table and do it separately.

Thank you!!

[Jai Ram Rideout]

Hi Neus,

Just as an example: Imagine that you have one OTU which is the only one present in a sample. Would it be removed if you apply a filtering of 1% and its fraction regarding the whole dataset is 0.6%?

Yes, that OTU would be removed because it contains 0.6% of the total sequences in the .biom file, and the minimum percentage for an OTU to be retained is 1%. As another example, if you supplied --min_count_fraction 0.5, OTUs would be filtered from the table that have sequence counts making up less than 50% of the total sequence count in the .biom file. Does that make sense?

Is there a script to filter OTUs taking into account the minimum count fraction of each sample but using the .biom table of the whole dataset? I mean, without need for filtering samples from the out table and do it separately.

Unfortunately I don't think there's an easy way to do this. Performing separate filtering like you described should work, and it may be possible to automate with some shell scripting.

Best,

Jai