Jan Axtner, Alex Crampton-Platt, Lisa A Hörig, Azlan Mohamed, Charles C Y Xu, Douglas W Yu & Andreas Wilting (2019) "An efficient and robust laboratory workflow and tetrapod database for larger scale environmental DNA studies." GigaScience 8(4):giz029. DOI: 10.1111/1365-2656.13555
The use of environmental DNA for species detection via metabarcoding is growing rapidly. We present a co-designed lab workflow and bioinformatic pipeline to mitigate the 2 most important risks of environmental DNA use: sample contamination and taxonomic misassignment. These risks arise from the need for polymerase chain reaction (PCR) amplification to detect the trace amounts of DNA combined with the necessity of using short target regions due to DNA degradation.
Our metabarcoding workflow should help research groups increase the robustness of their results and therefore facilitate wider use of environmental and invertebrate-derived DNA, which is turning into a valuable source of ecological and conservation information on tetrapods.
A series of bash and R scripts used in Axtner et al. 2019 "An efficient and robust laboratory workflow and tetrapod database for larger scale environmental DNA studies".
Provides a complete analysis pipeline for paired-end Illumina metabarcoding data, from processing of twin-tagged raw reads through to taxonomic classification with PROTAX.
The pipeline steps that will be of most use to other projects are those related to PROTAX: generating curated reference databases and associated taxonomic classification; training PROTAX models (weighted or unweighted); and classification of OTUs with PROTAX.
Steps and associated scripts:
- Process twin-tagged metabarcoding data
- read_preprocessing.sh
- Obtain initial taxonomic classification for target taxon
- get_taxonomy.sh
- Generate non-redundant curated reference sequence database for target amplicon(s) and fix taxonomic classification
- get_sequences.sh
- Train PROTAX models for target amplicon(s)
- train_protax.sh (unweighted) or train_weighted_protax.sh (weighted)
- check_protax_training.sh (makes bias-accuracy plots)
- Classify query sequences (reads or OTUs) with PROTAX
- protax_classify.sh or protax_classify_otus.sh (unweighted models)
- weighted_protax_classify.sh or weighted_protax_classify_otus.sh (weighted models)
Note: in some steps the screenforbio-mbc release associated with the manuscript is specific to the amplicons used in the study - primer sets and relevant settings are hard-coded in read_preprocessing.sh and get_sequences.sh. This will be generalised in a future release.
Pipeline tested on Mac OSX (10.13) and Scientific Linux release 6.9 (Carbon).
Mac OSX should have GNU grep, awk and sed prioritised over BSD versions. These can be installed with homebrew.
- Processing of raw reads only
- bcl2fastq (v2.18)
- AdapterRemoval (v2.1.7)
- cutadapt (v1.14)
- usearch (v8.1.1861)
- Building databases and PROTAX
- R (v3.5.0)
- tabtk (r19)
- seqtk (v1.2-r94 and v1.2-r102-dirty)
- seqkit (v0.7.2 and seqkit v0.8.0)
- Entrez Direct (v6.00 and v8.30)
- usearch (v8.1.1861 and v10.0.240)
- blast (v2.2.29 and v2.7.1)
- mafft (v7.305b and v7.313)
- sativa (v0.9-57-g8a99328)
- seqbuddy (v1.3.0)
- last (926)
- perl (v5.10.1 and v5.26.1)
get_sequences.sh also requires MIDORI databases for mitochondrial target genes Machida et al., 2017. Download relevant MIDORI_UNIQUE FASTAs in RDP format from the website. Manuscript used MIDORI_UNIQUE_1.1 versions of COI, Cytb, lrRNA and srRNA. Unzipped FASTAs should be placed in working directory.
get_sequences.sh also requires collapsetypes_v4.6.pl to be in the screenforbio-mbc directory. Download from Douglas Chesters' sourceforge page.
PROTAX scripts are reposted here with the kind permission of Panu Somervuo. These are in the protaxscripts subdirectory of screenforbio-mbc. This version of PROTAX is from Rodgers et al., 2017, scripts were originally posted on Dryad.
All steps in the pipeline are implemented via bash scripts with similar parameter requirements. Each script includes commented usage instructions at the start and displays the same instructions if run without any or an incorrect number of parameters.
Some of the bash scripts are primarily wrappers for R scripts, all of which are assumed to be in the screenforbio-mbc directory.
get_taxonomy.sh example:
mkdir ~/Documents/mbc_analysis
cd ~/Documents/mbc_analysis
bash ~/Documents/screenforbio-mbc/get_taxonomy.sh
You will see the following message:
You are trying to use get_taxonomy.sh but have not provided enough information.
Please check the following:
Usage: bash get_taxonomy.sh taxonName taxonRank screenforbio
Where:
taxonName is the scientific name of the target taxon e.g. Tetrapoda
taxonRank is the classification rank of the target taxon e.g. superclass
screenforbio is the path to the screenforbio-mbc directory
To get the ITIS classification for Mammalia:
bash ~/Documents/screenforbio-mbc/get_taxonomy.sh Mammalia class ~/Documents/screenforbio-mbc
When the script is running various messages will be printed to both the terminal and a log file. On completion of the script the final messages will indicate the next script that should be run and any actions the user should take beforehand.
Enjoy :-)