Hybrid-capture-enabled longitudinal metagenomics allows strain-resolved human-associated virus surveillance in wastewater

Thank you for your interest in our study!

This repository contains processed datasets and R scripts to reproduce the analyses and figures from the study. It also documents how the raw reads were processed so the workflow can be reproduced end-to-end.

Dataset (raw reads)

BioProject: PRJNA1391037 (SRA). Human-associated reads were removed before deposition.

Raw reads can be downloaded using the SRA Toolkit (e.g., prefetch / fasterq-dump). See: https://www.ncbi.nlm.nih.gov/sra/docs/sradownload/

Processed inputs for figure reproduction are included in data/ so you can regenerate figures without re-running the raw-read pipeline.

Metagenomics analyses (EsViritu v0.2.3)

Raw reads were processed with EsViritu v0.2.3 using the Virus Pathogen Database v2.0.2 (GenBank content through Nov 2022; Zenodo 7876309) as the reference. EsViritu links:

• GitHub: https://github.com/cmmr/EsViritu/
• Paper (Tisza et al., Nat Commun 2023): https://www.nature.com/articles/s41467-023-42064-1 If you use EsViritu, please cite the related paper.

Workflow summary:

• Quality filtering and adapter trimming with fastp, with deduplication enabled.
• Reference mapping at >=90% nucleotide identity and >=90% read coverage.
• Consensus sequence generation with samtools; near-duplicate consensus sequences removed at >95% similarity.
• The pipeline reports all findings (we have removed internal reporting cutoff of EsViritu); downstream filtering was done in RStudio with cutoffs mentioned below.

These steps produce the processed datasets in data/ that are used by the figure scripts.

Pre-filtering in R (create_filtered_data.R)

The pre-filtering step is implemented in scripts/create_filtered_data.R. It removes known wet-lab contaminants and low-confidence hits to improve certainty of detections and downstream genomic assignments. We also exclude a short-amplicon contamination in Human mastadenovirus C (<2 kb). Two filtered datasets are produced:

• Standard: >=1000 bp coverage or >=50% completeness, and >10 reads.
• Relaxed: >=500 bp coverage or >=50% completeness, and >10 reads, used for analyses where low-read targets would otherwise be lost; after contaminant removal and the read-count filter, the chance of spurious low-coverage hits is reduced. Whenever this is used, we explicitly noted it in the manuscript. Otherwise, we use the general cutoffs. You can also follow this up in the scripts.

Clinical datasets

Clinical datasets used in the study are included in this repository for reproducibility. These data sources are cited in the manuscript and are available under data/ (e.g., public_cases.csv, rva_weekly_cases.csv, uzleuven_pathogens_weekly_long.csv).

Rotavirus A genotyping

Rotavirus A segment genotypes were assigned by competitive mapping against RCWG genotype reference sets:

• Reference FASTA/TSV files per segment are stored in data/database_for_RVA_genotyping/.
• The competitive mapping workflow used influenza_a_serotype v0.1.5, replacing its influenza references with the RCWG segment references.
• For each segment, reads are aligned to all genotype references and the best-supported genotype is assigned; ambiguous reads are marked as ambiguous and are not shown in figures.

influenza_a_serotype links and related paper:

• GitHub: https://github.com/mtisza1/influenza_a_serotype
• Related paper (Tisza et al., N Engl J Med 2024): https://www.nejm.org/doi/full/10.1056/NEJMc2405937 If you use influenza_a_serotype, please cite the related paper.

Repository structure

.
├── data/                       # Processed datasets (RDS and CSV)
│   ├── filtered_data_wwlong.rds
│   ├── filtered_data_wwlong_500bp.rds
│   ├── processed_clinical_rotavirus.rds
│   ├── processed_wastewater_rotavirus.rds
│   ├── public_cases.csv
│   └── ...
├── scripts/                    # R scripts for generating figures
│   ├── Figure_1A.R             # Percentage of aligned reads
│   ├── Figure_1B.R             # Sequences per coverage category
│   ├── Figure_3ABCDEFGHI.R     # Rotavirus genotype analysis
│   └── ...
├── figures_pdf/                # Output figures in PDF format
└── README.md                   # Project documentation

Prerequisites

System requirements:

• Operating system: macOS (analysis was run on macOS)
• R version: >= 4.0.0
• No special hardware required.

R dependencies:

install.packages(c(
  "tidyverse",
  "lubridate",
  "ggplot2",
  "patchwork",
  "RColorBrewer",
  "gridExtra",
  "scales",
  "grid",
  "ggpubr",
  "ggrepel",
  "viridis",
  "zoo"
))

Usage (figure reproduction)

1. Clone the repository:

   git clone https://github.com/Matthijnssenslab/Leuven_ww_long.git
   cd Leuven_ww_long

2. Set the working directory in R/RStudio to the repository root:

   setwd("/path/to/Leuven_ww_long")

3. Run scripts named after the figures they generate, for example:

   source("scripts/Figure_1A.R")

Note: Some figure panels were generated separately and assembled in Adobe Illustrator 2024 without editing data-associated elements. Virus taxonomy labels were updated to current ICTV naming. Run time is <1 min for each figure.

Data description

• filtered_data_wwlong.rds: Main dataset containing filtered sequencing reads.
• processed_clinical_rotavirus.rds: Anonymized clinical surveillance data for rotavirus comparisons.
• processed_wastewater_rotavirus.rds: Aggregated wastewater data for rotavirus genotyping.

License

MIT License. See LICENSE.

Citation

If you use this code or data, please cite:

Mustafa Karatas, Mandy Bloemen, Jill Swinnen, Lila Close, Marc Van Ranst, Elke Wollants, Jelle Matthijnssens. "Hybrid-capture-enabled longitudinal metagenomics allows strain-resolved human-associated virus surveillance in wastewater". SSRN Preprint. DOI: https://doi.org/10.2139/ssrn.6047769