eDNA metabarcoding workflow

An adaptable workflow for processing and analyzing multi-marker environmental DNA metabarcoding data. This repository documents the processing and analysis steps used in a study combining COI, 18S rRNA, and 16S rRNA assays.

The workflow covers:

• marker-specific APSCALE processing
• BLASTn searches and lowest-common-ancestor taxonomic reconciliation
• control-based contaminant subtraction at the PCR-replicate level
• PCR-replicate, filter, and sampling-event aggregation
• marker-specific relative read abundance filtering
• multi-marker presence-absence integration
• richness, accumulation, ordination, PERMANOVA, LCBD, indicator-taxon, and co-detection network analyses

This repository is an analysis companion rather than a one-command software package. File paths, metadata values, control scopes, and manually reviewed taxonomy must be adapted to each project.

Workflow overview

paired-end FASTQ files
        |
        v
primer handling and APSCALE processing (COI, 18S, and 16S separately)
        |
        v
ESV occurrence tables
        |
        v
BLASTn searches and NCBI lineage retrieval
        |
        v
candidate-hit filtering and LCA assignment
        |
        v
control subtraction and PCR-replicate aggregation
        |
        v
manual taxonomy review and event-level aggregation
        |
        v
marker-specific filtering and multi-marker matrices
        |
        v
community, sampling-effort, indicator, and network analyses

Repository structure

config/      marker thresholds and project-wide settings
data/        metadata and reviewed-taxonomy templates
docs/        run order, input schemas, and methodological notes
hpc/         generic SLURM templates for APSCALE, BLASTn, and COI trimming
R/           reusable helper functions
scripts/     ordered R workflow stages
results/     generated outputs; contents ignored by Git
workflow_template.R

Quick start

1. Clone or download the repository.
2. Create the local input and output directories listed in config/settings.R.
3. Copy and complete the templates in data/metadata/.
4. Adapt the paths and scheduler settings in hpc/ before running the marker-specific preprocessing jobs.
5. Copy workflow_template.R, populate the input file paths, and run the R scripts in numerical order.
6. Review LCA assignments and save the accepted tables under data/reviewed_taxonomy/ before building final ecological matrices.

Detailed instructions are provided in docs/RUN_ORDER.md.

Marker-specific identity thresholds

Marker	Species	Genus	Family	Order
COI	>=97%	>=94%	>=90%	>=85%
16S	>=98%	>=94%	>=90%	not assigned by identity threshold
18S	>=97%	>=92%	>=88%	>=85%

Candidate hits must also agree taxonomically. A high percentage identity alone does not guarantee a species-level assignment. Thresholds are stored in config/marker_config.csv.

Multi-marker integration

Read abundance is calculated and filtered independently within each marker. Because read proportions are not directly comparable among assays, the default cross-marker community matrix is presence-absence. The repository retains a separate marker-level RRA table for marker-specific summaries and plots.

Required software

• R 4.3 or later
  • metabark
  • vegan
  • indicspecies
  • adespatial
  • igraph
  • ggraph
  • ggplot2
• cutadapt
• APSCALE
• BLAST+
• A local NCBI nucleotide database or another documented reference database

Required R packages are checked in scripts/00_packages.R. The optional online NCBI lineage-retrieval step additionally requires metabark and an NCBI API key stored locally as NCBI_API_KEY in .Renviron. No credentials are stored in this repository.

Reproducibility and data availability

Raw FASTQ files, BLAST databases, large intermediate tables, manually reviewed project data, and generated results are intentionally excluded from Git. Public data accessions and the associated manuscript can be added when available.

Before creating a release, record package versions with renv or save the output of sessionInfo().

License

Code is released under the MIT License.