Issue with creating biom file (make.biom) for use in FAPROTAX protocol

[parjker]

Hello all. I am attempting to do functional assignment for my microbial samples. I have already clustered OTUs and done Taxonomic assignment. These are the last commands I used to create my main output files (i.e. my .shared and .cons.taxonomy files):

cluster.split(fasta=final.fasta, count=final.count_table, taxonomy=final.taxonomy, taxlevel=4, cutoff=0.03); make.shared(list=final.opti_mcc.list, count=final.count_table, label=0.03); classify.otu(list=final.opti_mcc.list, count=final.count_table, taxonomy=final.taxonomy, label=0.03

I want to use FAPROTAX, which says:

Input OTU/taxon tables can either be in classical format (e.g. with tab-separated columns, one column per sample and one row per OTU), or in the recently published BIOM format (JSON or HDF5 variants)

Since mothur doesn’t output files in the classical format, I’m attempting to convert my file into .biom using the make.biom command.

Here is what I code into mothur (using a supercomputer on campus):

mothur "#set.dir(input=/projects/cas010/bartz/16S_Files, output=/projects/cas010/bartz/16S_Files/biom);
make.biom(inputdir=/projects/cas010/bartz, shared=final.opti_mcc.shared, constaxonomy=final.opti_mcc.0.03.cons.taxonomy)"

or, I’ve also tried:

mothur "#set.dir(input=/projects/cas010/bartz/16S_Files, output=/projects/cas010/bartz/16S_Files/biom2);
make.biom(inputdir=/projects/cas010/bartz/16S_Files, shared=final.opti_mcc.shared, label=0.03, reftaxonomy=trainset9_032012.pds.tax, constaxonomy=final.opti_mcc.0.03.cons.taxonomy)"

Both of these run very quickly and give me the same output. Although I’m not sure what a .biom file is supposed to look like exactly, I think this output .biom file looks strange. Here is the beginning of the file, opened in TextEdit, for example:

I can’t use this table in the FAPROTAX protocol. I can’t read this into R using read_biom in the “biomformat” package. And I can’t get a summary using summarize-table in the terminal. It runs a while and in the end says KeyError: “Unable to open object (object ‘group-metadata’ doesn’t exist)”

What am I doing wrong here? Any advice is much appreciated! Thank you. If you need any more details please let me know.

[pschloss]

Hi - to be honest, I know very little about biom formatted files and we generally don’t use them unless someone else already produced the biom file. Ours is outputted as a HDF5-formatted file, which is a binary format. This means you can’t read it (and make anything out of it) in a text editor. I seem to recall that the biomformat package functions have an argument to change the format. I think the default is to read the classical json format rather than the HDF5 format.

It may also be that you need to use the approach we describe in the make.biom wiki page for using the files with picrust.

Sorry I’m not able to be more helpful
Pat