Error: Unable to create link (Name already exists)

1,230 views
Skip to first unread message

Samuel Major

unread,
May 3, 2017, 9:17:37 AM5/3/17
to Qiime 1 Forum
Hi everyone, 
I'm trying to create my otu table. I picked otus with usearch61, picked my rep set and then assigned taxonomy.
now when I run:
 
      make_otu_table.py -i 3_usearch_qf_results/combined_seqs_otus.txt -t MSC/3_usearch_qf_results/uclust_assigned_taxonomy/rep_set1_tax_assignments.txt -m MSC_map.txt -o 3_usearch_qf_results/MSC_otu1.biom

I still get an otu table, but I also get an error:
      RuntimeError: Unable to create link (Name already exists)

When i try to summarize this biom table, that's when it told me it's that this biom file is not really a biom file at all!

Can anyone help me with this?
thank you very much
Sam

Greg Caporaso

unread,
May 3, 2017, 2:30:14 PM5/3/17
to Qiime 1 Forum
Hi Sam,
Based on a previous user who ran into this same issue, I think the BIOM hdf5 format may not be able to support some of the sample metadata that you're providing. You have a couple of options here. 

Option 1: You can just not include the -m MSC_map.txt option to make_otu_table.py. QIIME doesn't ever use the sample metadata from the biom file, so there really isn't any benefit to including it. Any time the sample metadata is used by QIIME, you're required to pass the sample metadata mapping file. In this case, your new command would look like:

make_otu_table.py -i 3_usearch_qf_results/combined_seqs_otus.txt -t MSC/3_usearch_qf_results/uclust_assigned_taxonomy/rep_set1_tax_assignments.txt -o 3_usearch_qf_results/MSC_otu1.biom

Option 2: If there is a specific reason why you want to include the sample metadata in the biom file, you can try uninstalling h5py. This will cause QIIME to fallback to writing BIOM files in the JSON format, and should get you around this error.  

Can you follow up here to let us know if one of these two options ends up working for you? It'll be helpful for us and other users to have that information. 

Best,
Greg

Samuel Major

unread,
May 3, 2017, 7:05:35 PM5/3/17
to Qiime 1 Forum
Great! dropping the metadata mapping file seemed to do the trick. 
Will this have any effect on separating samples from each other downstream?

Thank you
Sam

Greg Caporaso

unread,
May 4, 2017, 2:29:08 PM5/4/17
to Qiime 1 Forum
Great, glad it worked! Thanks for following up.

QIIME doesn't use the sample metadata from the BIOM table (it always has you provide the sample metadata independently when it's used), so this won't impact any downstream analyses. 

Best, 
Greg
Reply all
Reply to author
Forward
0 new messages