Rhea is a software used to detect structural variants (SVs) between steps in long-read metagenomic series data. Current SVs detected include: insertions, deletions, complex indel (insertion and deletion in the same location), and tandem duplications. Below is a graphic represtation of the rhea pipeline.

Here is a toy example to detect SV in two different variants E. coli in a metagenome. First, download t0.fasta and t1.fasta from OSF under the example directory. Then run rhea on the samples.

python ./rhea.py example/t0.fasta example/t1.fasta

Expected output: All output files are expected to be in directory rhea_results. Output files include: Bandage_metadata.csv, bp_counts.tsv, edge_coverage-t{0,1}, metaflye directory, node_coverage_norm.csv, node_coverage.csv, rhea.log, structual_variants-c0, t{0,1}.gaf.

Expected run time: ~5 minutes for flye

Dependencies for rhea include: python v3.8+, numpy, pandas, networkx v3.2+, seaborn, Flye, and minigraph. Environment.yaml can be used build a conda environment with all necessary dependencies; rhea environment will need to be activated for each rhea run.

https://github.com/treangenlab/rhea.git
cd rhea
conda env create -f environment.yml
conda activate rhea

We highly recommend using metaFlye v2.9.3+ due to improvements in strain variations maintained.

• structual_variants-c{1-N}.tsv: detected structual variants for each pair of subsequent samples. Denoted in the file name.
• node_coverage.csv: Spreadsheet where each row is a unique node. Contains sequence length for each node, detected coverage for each sample, linear change in coverage between subsequent pairs, and log fold change in coverage. Color hex values are also provided for each type of coverage/coverage difference for graph heatmap visuals.
• node_coverage_norm.csv: Same as node_coverage.csv, except coverage values are normalized across all samples in the set based on total number of base pairs per sample. Use of normalized or raw values for SV detection is determined by the input parameter --raw-diff (normalized is default).
• edge_coverage-c{}.tsv: 2D matrix of edge coverage for each provided sample.
• Bandage_metadata.csv: consolidated data for regarding node length, coverage, coverage change, and detected SVs; designed to be used with Bandage to create visuals. See below.
• bp_counts.tsv: total number of bps in each sample; used for normalization.
• rhea.log: Computational log and established min/max values for heatmap graph visuals.

Rhea output can be used in conjunction with Bandage to see patterns SV and evolutionary activity in the metagenome.

1. Install and open Bandage.
2. Load assembly graph: ./rhea_results/metaflye/assembly_graph.gfa.
3. Select column for node colors from Bandage_metadata.csv by changing any of the column name ending in _colour to colour.
4. Import CSV for updated CSV into Bandage.
5. Draw graph single with custom colours.
6. Min/max color scale values can be found in rhea.log. Colored bar can be found in docs/visual_scales.png.

If you cite rhea, be sure to also cite metaFlye and mingraph. If you use rhea visuals, be sure to cite Bandage.

Rhea preprint: Curry, K. D., Yu, F. B., Vance, S. E., Segarra, S., Bhaya, D., Chikhi, R., Rocha, E. P. C., & Treangen, T. J. (2024). Reference-free Structural Variant Detection in Microbiomes via Long-read Coassembly Graphs (p. 2024.01.25.577285). bioRxiv. https://doi.org/10.1101/2024.01.25.577285

Validation tests & scripts: https://osf.io/fvhw8/