alpha_diversity vs alpha_rarefaction

[Ricardo Ramiro]

Hi all,

I ran the core_diversity_analysis script and obtained values of observed OTUs on the collated alpha diversity folder. However, as I wanted to get a single value per sample I then ran the alpha_diversity script on a single biom table. Now in the collated file from the rarefaction the maximum sampling depth is 21730 and the biom table is done at 21735 for all samples. At the 21730 depth, the number of observed OTUs in the collated table is around 500-700 but this is 2 to three times higher in the table generated by the alpha_diversity script. I guess I could understand big discrepancies like this if the difference between the two sampling depths was large, but here that is not the case. So if anyone can help me figure out what might be going on that would be great.

Thanks,

Ramiro

[abir...@gmail.com]

Hi, Ramiro,

Could you please provide the command lines that you ran and the OTU tables they produced?  This will help us understand what's going on!

Thanks,

Amanda

[Ricardo Ramiro]

Hi Amanda,

Thanks for your reply. The commands were the following:

alpha_diversity.py -i table.biom -m chao1,PD_whole_tree,observed_otus,simpson,simpson_e,shannon -o alpha_div_metrics.txt -t rep_set.tre

core_diversity_analyses.py -o cdout_21735 -c group -i table.biom -m mapping.tab -t rep_set.tre -e 21735

[abir...@gmail.com]

Hi, Ramiro,

In the alpha_diversity.py command above, what is the source of the table.biom file that you are using?  Please provide what info you can about how it was generated (including the command, if possible) ... based on the limited details here, I'm wondering if perhaps this is an un-rarefied OTU table.

Thanks,

Amanda

[Ricardo Ramiro]

Hi Amanda,

so after mapping the reads to OTUs, I used the following command to generate the table.biom:

filter_samples_from_otu_table.py -i otu_table_mc2_w_tax_no_pynast_failures.biom -o table.biom -m samples.tab -s 'group:blue,red_high' -n 21735

sorry for not sending all the details from the start, i am also unsure which info will help.

Thanks,

Ramiro

[abir...@gmail.com]

Hi, Ramiro,

Ok, I think this helps.   In your first post, you mentioned "the biom table is done at 21735 for all samples".  If you are referring to the use of the -n 21735 switch in your filter_samples_from_otu_table.py command, I think this may be the source of the confusion.  That switch does NOT rarefy the OTU table to a depth of 21735, but only filters out of the OTU table any samples with fewer than 21735 total observations (see http://qiime.org/scripts/filter_samples_from_otu_table.html for more details).  I'd guess this is probably why you saw so many more OTUs in this table than in the rarefied one.

Best,

Amanda

[Ricardo Ramiro]

Hi Amanda,

Thanks a lot for your reply, I guess I had misinterpreted the "filter" command. I just have one final question:

- I would like to have a single value of alpha diversity per sample. I was thinking that I could go to the highest sampling depth in the collated tables of the alpha diversity and take the mean of the 10 different runs. Would it be ok to do this? or would it be better to just take a random value from this set of 10?

kind regards,

Ricardo Ramiro