PICRUST does not recognize my file

Continuing the discussion from How to create a feature table with qiime2 for PICRUST with the taxonomic assignment?:

I followed all the steps mentioned in this post but when I try to analyze using Galaxy, it cannot recognize my file as a PICRUST format file. I used my feature-table from dada2 to do a close-reference alignment and then converted that to biom file and merged with taxonomy file. I am not sure why it is not working. How large should a biom file be? Mine is pretty small like about 500kb.

Here are the codes I used:

Make close-reference files for PICRUST:

  1. Download greengenes reference version gg-13-5-OTU from 2013 that picrust is trained with
    gg_13_5_otus.tar.gz (download)

  2. Import greengenes reference version gg-13-5-OTU from 2013 that picrust is trained with
    qiime tools import
    –type ‘FeatureData[Sequence]’
    –input-path Aligned_close/gg-13-5-97_otus.fasta
    –output-path Aligned_close/gg-13-5-97_otus.qza

  3. close-reference alignment
    qiime vsearch cluster-features-closed-reference
    –i-sequences …/qza/v3v4-20190304-rep-seqs-dada2-p-Phylum-only-nNC.qza
    –i-table …/qza/v3v4-20190304-table-dada2-p-Phylum-nNC.qza
    –i-reference-sequences Aligned_close/gg-13-5-97_otus.qza
    –p-perc-identity 0.97
    –p-strand both
    –output-dir Aligned_close/aligned

  4. Export the feature table
    qiime tools export
    –input-path Aligned_close/aligned/clustered_table.qza
    –output-path Aligned_close/exported

  5. add the corresponding Greengenes taxonomic annotations using biom add-metadata
    biom add-metadata -i Aligned_close/exported/feature-table.biom -o Aligned_close/exported/table-with-taxonomy.biom --observation-metadata-fp Aligned_close/97_otu_taxonomy.txt --sc-separated taxonomy

Hi @Negin,

I suspect you’ve got a biom incompatibility problem. I think the Galaxy 1 server assumes a biom 1 format and the standard export from QIIME 2 is biom 2. I think you can export the table as biom 1 if you look at qiime tools export.

In the mean-time, have you tried getting the QIIME 2 PICRUSt 2 plugin to work? Its more computationally expensive, but it does have the advantage of a new reference and it works with your ASVs.


Hi Justine,

I tried QIIME 2 PICRUSt 2 and it worked but it is does not give me info on which pathways belong to which samples so I am not sure what I can do with what I got from that. I do not like to run diversity analysis with kegg pathways. I would like to be able to see which sample is showing what specific kegg pathways but I can’t see that using csv files I export from the results. I either get sequence counts (which I imagine number of sequences related to specific pathways?) and frequency of pathways and # of samples those pathways are observed in. What I would like to know, however, is to see which one of these pathways are related to which of my samples.

I tried installing picrust on my computer and running through there. It keeps telling me that the input file does not exist even though the path is correct.
normalize_by_copy_number.py: error: option -i: file does not exist: ‘–input_otu_fp’

I cannot find info about conversion to different biom files in qiime.

Hi @Negin,
With your work now being completely separated from Qiime2 (i.e. the native PICRUSt installation) I think you will have better luck getting your questions answered on the PICRUSt Forum.

As for converting your biom file to an older version, in the past I’ve just just basically converted Biom2 to text and reconverted back to Biom1. There’s probably a more reasonable way to do this but hey it worked for me. I’ve put a little mini instruction on how to do this here. I made it for BugBase but it is compatible with PICRUSt since that’s the underlying tool in BugBase anyways.


Hi Merhbod,

So cool! Thanks! I will try that.

Conversion didn’t help either. Here are all the files:

[I removed the files.]

I did post my question in picrust google forum so let’s see whether I can get any help from there.


Hi @Negin,

Sorry Im back late.

I think that count table is what you want. When I run PICRUSt, I get out three feature tables: an ec table, I think a ko table and a pathway table which is the EC table collapsed. It doesn’t currently have the functionality to tell you which ASVs contributed ot which pathway directly, but you still should get the feature table?


Hi Justine,

I got all the four different tables:




I would like to know which host samples (not microbes) are related to these pathways. This gives me no info for comparing the host species I have othe than comparing beta and alpha diversity which I am not sure would be the best way of analyzing pathways.

In the PICRUSt tutorial, using ANCOM is discussed. Like I said, you get a feature table of compositional data out of PICRUSt. You can do with it whatever you’d do with a feature table of compositional data without a phylogenetic tree.

That said, I tend to find beta diversity pretty useful, but Im of the school that if you don’t have a difference in beta diversity you shouldn’t do feature-based analysis because I prefer type II error to type I…


1 Like

This topic was automatically closed 31 days after the last reply. New replies are no longer allowed.