how to build a db from multiple input files and directories?

[rjsorr]

Hi,

I have multiple seperated downloads from genome_updater, that I want to keep seperate. For example I have an Archaea, Bacteria and Fungus genome_updater directrory. Now I want to make a Taxor database of all 3 together, but how do I do this? Can I point to all three by simply using "*" in the path or can I, like kraken, add one after the other?

regards

[JensUweUlrich]

@rjsorr
Taxor currently cannot deal with different directories of genome_updater. I'm actually thinking about Taxor accepting comma-separated lists of input directories (for the reference genome fasta files) and input files (for the taxonomic information). Is this something that would help you?

[ulrikhorn]

Hello,
i am interested in the same as @rjsorr and a csv list of input directories sounds like it would work just fine. As long as it is possible to merge different downloads into a single taxor database.
Thank you!

[eh58]

would symlinks work?

[eh58]

would symlinks work?

I just tried and it doesn't work.

I have all my fasta files stored in subdir as archaea, bacteria, viral, etc. All other tools can be built with this file structure but I am having trouble with taxor now. Please let me know if I will have to make a copy of all the files to a single directory to build taxor. It's 200+GB so I'd try my best to avoid doing that if possible.

Thank you!

[JensUweUlrich]

Hi,

symlinks unfortunately do not work at the moment. I have to investigate what I will need to change to make it possible. In the meantime, the latest taxor release (v0.2.0) has a new feature to provide several different directory paths as a comma-separated list, which makes copying of files into one single input directory obsolete. I hope this helps a bit and may solve your issue.

[eh58]

Hi,

symlinks unfortunately do not work at the moment. I have to investigate what I will need to change to make it possible. In the meantime, the latest taxor release (v0.2.0) has a new feature to provide several different directory paths as a comma-separated list, which makes copying of files into one single input directory obsolete. I hope this helps a bit and may solve your issue.

Great! That would be really helpful!
I can access the latest release by clone the package locally but I am wondering if it's also available through conda?
I checked the conda site and it shows it's updated to v0.2.0 but after executing conda update taxor, it remains at v0.1.3. Did I do anything wrong?

Thanks!

[JensUweUlrich]

You can try conda install taxor=0.2.0 -c bioconda -c conda-forge. This worked for me.

[eh58]

You can try conda install taxor=0.2.0 -c bioconda -c conda-forge. This worked for me.

Yes! Thank you. I actually just resolved the issue

[eh58]

Should command like this work?
taxor build --input-file refseq_accessions_taxonomy.csv --input-sequence-dir library_gz/archaea/,library_gz/bacteria/,library_gz/fungi/,library_gz/viral/ --output-filename ./taxor_abfv_030325/refseq-abfv-k22-s12.hixf --threads 32 --kmer-size 22 --syncmer-size 12 --use-syncmer

I ran into the following error:

checking input ... terminate called after throwing an instance of 'std::runtime_error'
what(): No file name found forGCF_036169615.1 !!!
Aborted (core dumped)
--output-filename: command not found

[JensUweUlrich]

Taxor complains that there is no file name GCF_036169615.1*.fna.gz in your input directories. Besides that, there is an error in the command in front of the --output-filename parameter. Looks like too many blanks or a linebreak

[eh58]

Taxor complains that there is no file name GCF_036169615.1*.fna.gz in your input directories. Besides that, there is an error in the command in front of the --output-filename parameter. Looks like too many blanks or a linebreak

The blank is probably just a copy-paste issue.

So I think the file not found error is due to that the fasta files in my directory are less than the count in the generated refseq_accessions_taxonomy.csv file. The file was generated by following the instructions from the repo, but my customized database does not include every fasta file that appears in the assembly_summary.

For my study, I'd not change the tools' source code, so is it possible you can add exceptions to skip IDs if files are not found? I believe having a selected database is very common.

[JensUweUlrich]

Valid point. I will try to implement this change request asap. I'll let you know when I'm ready.