Half of reads fail to be placed in otu table during open reference OTU picking

[bala krish]

Hi,

It is bit strange that the pick_open_reference_otus.py script does not place all my reads in the otu table when i run the command with all my 900 samples. However, it works well and place almost all reads when i try with small number of samples. Is there any bug which could not deal with larger number of datasets. But in both case, the command executed sucessfully but the results were not correct for the fasta file with all the samples. I like to send the fasta file to reproduce the issue. So, could you send me an email requesting the file?

Thanks in advance!

Bala  

[Jai Ram Rideout]

Hi Bala,

I've contacted a developer who can help answer your question -- they will get back to you shortly.

Best,

Jai

[Greg Caporaso]

Hi Bala,

This might be a result of the --min_otu_size parameter, which defaults to 2 (meaning that OTUs that are only observed one time are filtered out). It also could result from some sequences being filtering if they are unalignable with PyNAST. You could test this by passing --min_otu_size 1, and then looking at the otu_table_mc1_w_tax.biom rather than the otu_table_mc1_w_tax_no_pynast_failures.biom file that is generated (the former doesn't have the unalignable sequences filtered). If you are losing reads due to these filters, they're probably reads that you don't want. If you're still finding that you're losing reads, let me know and we can try to figure out what else might be causing this. 

Hope this helps! 

Greg