Summary

Methods

hundo wraps a number of functions and its exact steps are dependent upon a given user's configuration.

Paired-end sequence reads are quality trimmed and can be trimmed of adapters and filtered of contaminant sequences using BBDuk2 of the BBTools package. Passing reads are merged using VSEARCH then aggregated into a single FASTA file with headers describing the origin and count of the sequence.

Prior to creating clusters, reads are filtered again based on expected error rates:

To create clusters, the aggregated, merged reads are dereplicated to remove singletons by default using VSEARCH. Sequences are preclustered into centroids using VSEARCH to accelerate chimera filtering. Chimera filtering is completed in two steps: de novo and then reference based. The reference by default is the entire annotation database. Following chimera filtering, sequences are placed into clusters using distance-based, greedy cluster with VSEARCH based on the allowable percent difference of the configuration.

After OTU sequences have been determined, BLAST or VSEARCH is used to align sequences to the reference database. Reference databases for 16S were curated by the CREST team and hundo incorporates the CREST LCA method. ITS databases are maintained by UNITE.

Counts are assigned to OTUs using the global alignment method of VSEARCH, which outputs the final OTU table as a tab-delimited text file. The Biom command line tool is used to convert the tab table to biom.

Multiple alignment of sequences is completed using MAFFT. A tree based on the aligned sequences is built using FastTree2.

This workflow is built using Snakemake and makes use of Bioconda to install its dependencies.

Configuration

fastq_dir: /Users/brow015/devel/hundo/example/mothur_sop_data
filter_adapters: None
filter_contaminants: None
allowable_kmer_mismatches: 1
reference_kmer_match_length: 27
reduced_kmer_min: 8
minimum_passing_read_length: 100
minimum_base_quality: 10
minimum_merge_length: 150
fastq_allowmergestagger: False
fastq_maxdiffs: 5
fastq_minovlen: 16
maximum_expected_error: 1.0
reference_chimera_filter: True
minimum_sequence_abundance: 2
percent_of_allowable_difference: 3.0
reference_database: silva
aligner: blast
blast_minimum_bitscore: 125
blast_top_fraction: 0.95
read_identity_requirement: 0.97

Execution Environment

name: hundo_env
channels:
    - bioconda
    - conda-forge
    - defaults
dependencies:
    - python>=3.6
    - bbmap=37.17
    - biom-format=2.1.6
    - biopython
    - blast=2.6.0
    - bzip2=1.0.6
    - click=6.7
    - docutils=0.14
    - mafft=7.313
    - fasttree=2.1.10
    - numpy>=1.14.0
    - pandas>=0.23.0
    - pigz=2.3.4
    - pyyaml>=3.12
    - vsearch=2.6.0
    - zip=3.0
    - pip
    - pip:
      - relatively
      - plotly>=3.0

Output Files

Not all files written by the workflow are contained within the Downloads section of this page to minimize the size of the this document. Other output files are described and are written to the results directory.

Downloads