16S rRNA gene sequence curation and phylogenetic reference set creation

• confirm availability of necessary libraries to compile dependencies (on Ubuntu: sudo apt-get install gfortran libopenblas-dev liblapack-dev)
• Install Python 2.7
• run bin/bootstrap.sh
• run source deenurp-env/bin/activate

the deenurp executable should now be on your $PATH

See required system libraries above.

First, install binary dependencies:

• Python 2.7

  • pip, for installing python dependencies (http://www.pip-installer.org/)

  • Python packages:

    • Run pip install PACKAGE for every PACKAGE listed in requirements.txt, e.g. cat requirements.txt | xargs -n 1 pip install

• vsearch (https://github.com/torognes/vsearch)

• Infernal version 1.1 (http://infernal.janelia.org/)

• pplacer suite (http://matsen.fhcrc.org/pplacer)

• FastTree 2 (http://www.microbesonline.org/fasttree/#Install)

Optional (for filter-outliers and pairwise-distances):

• muscle (http://www.drive5.com/muscle/)

Finally, install:

python setup.py install

Similarity-search based reference sequence selection

The deenurp package under the current directory provides to subcommands, accessed via the script deenurp.py, or the command deenurp if installed.

Subcommands fall into two general categories:

• Building a set of reference sequences for use in refpkg building
• Selecting sequences for a specific reference package

Removes outlier sequences from a reference database

Expands poorly-represented names in a sequence file by similarity search

Cluster reference sequences, first by tax-id at a specified rank (default: species), then by similarity for unnamed sequences or sequences not classified to the desired rank. Serves as input to search-sequences.

Builds a set of hierarchical reference packages.

Searches a set of sequences against a FASTA file containing possible reference sequences.

This subcommand does searches sequences against a reference FASTA file, saving the results and some metadata to a sqlite database for use in select-references

Given the output of search-sequences, select-references attempts to find a good set of reference sequences.

For each reference cluster with a minimal amount of sequences having best hits to the cluster, (see cluster-refs), selects a set number of sequences to serve as references.

Taxa who are the sole descendent of their parent can complicate taxonomic classification.

The fill-lonely subcommand finds some company for these lonely taxa.

Fetches sequences from a sequence file which match the taxtable for a reference set at a given rank. Useful for adding type strains.

Runs the tax2tree program on a reference package, updating the seq_info file.

Sequences whose lineage changes are relabeled. The prior tax_id is added to the seq_info file in the reference package.