filter singletons doesn't work

[Irelka Colina]

Hello!

I used 2 commands in QIIME to prepare and to suprime singleton from samples.
When I saw the new file, I perceived some singletons in some samples even if the file was considerably reduced.

If someone can help me to undertand why this happen ? or if I need something more effective or if it is normal...

Thanks,
Irelka


These are the commands that I used,

make_otu_table.py -i uclust_picked_otus/XXX_ChimeraChecked_otus.txt -t uclust_assigned_taxonomy/rep_set_tax_assignments.txt -o otu_table.biom

filter_otus_from_otu_table.py -i otu_table.biom -o otu_table_no_singletons.biom -n 2

[Andrew Krohn]

This script filters singletons across the entire table, and not on a per-sample basis which explains your result.

Bokulich et al 2013 make the suggestion to filter across the entire table at 0.005% (http://www.ncbi.nlm.nih.gov/pmc/articles/PMC3531572/) rather than singleton filtering. The command would be:

filter_otus_from_otu_table.py -i otu_table.biom -o 005_table.biom --min_count_fraction 0.00005

[Irelka Colina]

Thanks Andrew !!!

I will try and if it is run I will tell you.

Irelka

[Irelka Colina]

Hello!

I used your command and it is run !!!

Thank you !!!

[Andrew Krohn]

Glad that it worked. I wrote up a script this morning using a colleague's python creation as the back end to filter OTUs by sample rather than by table. My colleague's script works well, but the documentation wasn't great. I haven't tested this script much yet, so use at your own risk. Capable of filtering singletons by sample, or specifying a decimal value of per sample abundance to filter. You will need my github repository to make it work .

Script wiki page: https://github.com/alk224/akutils-v1.2/wiki/filter_observations_by_sample.sh

akutils repository: http://alk224.github.io/akutils-v1.2/

[Irelka Colina]

Hello !
Thank you, for yoyr help ! Maybe I wil use but I not sure because I have 64 samples ;)

Now, I have a problem with my new file (filtered 005%), when I try to do a single rarefaction in RStudio because I need it for a multiple rarefaction it does not work :'( but it works with my file that was filtered with first command that I wrote you.
I would know if you have an idea about that ?

I send you the script that I used if you want see it...

Muchas Gracias,
Irelka

[Andrew Krohn]

I'm not great at R, and I only use it when I have to. There is a rarefaction command in QIIME which will do what you need. If you insist on R, likely you will have to spend some time making your file properly formatted.

Try these out for QIIME use:

http://qiime.org/scripts/single_rarefaction.html

http://qiime.org/scripts/multiple_rarefactions.html

http://qiime.org/scripts/alpha_rarefaction.html