Issues with summarize_taxa.py and "ValueError: cannot convert float NaN to integer" Error

[Nicholas Ellaby]

Dear All, 

New to the forum, not so new to QIIME, that being said I'm not as tech savvy as I'd wish and I was wondering if someone would be able to point my in the right direction as to where/why I'm encountering the error I am.

############

Bit of background, I've performed 13 MiSeq 2x250 pe sequence runs, ~1200 samples all in all.

Each run was analysed with the exact same pipeline script I have encountered errors on.

Reads were error corrected with Spades Bayes Hammer first

Overlapped with PandaSeq

Re-named with perl script.*

Concatenated all samples into single '.fasta' file.

Then ran through the de novo QIIME pipeline using usearch61** 

* - naming syntax slightly different, instead of 

>CN10_10013 orig_bc=CGAGGCTG-AGAGTAGA

Changed to include sequence run in name:

>SR10=CN10_10013 orig_bc=CGAGGCTG-AGAGTAGA

** - Used uclust for individual runs, however, after scaling up the commercial version was used.

To note; no error messages at all were encountered in the pipeline up to the point of the error

I've attached the pipeline script for reference, un-hashed section is were errors are being observed.

############

Error:

When using the script:

summarize_taxa.py -i OTUS/20_sorted_otu_table_0_05.biom -o taxa_summaries_test_L1-7_relative -L 1,2,3,4,5,6,7 

It returns the following error

/usr/local/share/qiime180/biom-format-1.3.1-release/lib/python2.7/site-packages/biom/table.py:1152: RuntimeWarning: invalid value encountered in true_divide

  return samp_v / float(samp_v.sum())

Now I am really unsure as to the root of this problem and any advice would be greatly appreciated.

Sincerely, 

nick