Skip to content

SOP: Fish Mitogenome Assembly and Annotation from ONT Data

Version: 1.1 Date: 2026-03-30 Pipeline scripts: run_pipeline.sh, annotate_mitos2.sh, generate_report.sh

1. Overview

This SOP describes how to assemble and annotate a fish mitochondrial genome (mitogenome) from Oxford Nanopore Technology (ONT) long reads. The pipeline takes raw reads, enriches mitochondrial reads by mapping to a reference, assembles and polishes the mitogenome, annotates it using two complementary tools (MITOS2 for gene annotation and MitoZ for circos visualisation), performs BLAST-based species identification against NCBI nt, and generates a final markdown report.

The pipeline is configured per species using a plain-text config file, allowing it to be reused across any fish species without modifying the main scripts.

Expected outputs for a typical vertebrate fish mitogenome:

  • 1 circular contig, ~16–18 kb
  • 13 protein-coding genes
  • 2 ribosomal RNA genes (12S, 16S)
  • 22 transfer RNA genes

2. Pipeline Code Repository

The pipeline scripts are version-controlled in a dedicated Gitea repository:

172.20.142.126:3000/evilliers/mitogenome-pipeline

This repository contains:

File / Folder Description
run_pipeline.sh Main assembly pipeline (Steps 0–8)
annotate_mitos2.sh MITOS2 annotation (run separately)
generate_report.sh Final report generator
configs/ Species config files (one per species)
SOP_mitogenome_pipeline.md This SOP (kept in sync with the website)

Large files excluded from git

The following are not tracked in git due to size: singularity/mitoz_3.6.sif (~2 GB), results/, reference/, logs/, and raw read files. These are generated or downloaded automatically on first run.

Contributing changes via pull request

If you update any pipeline script or config file, please submit a pull request rather than pushing directly to main:

# 1. Clone the repo (first time only)
git clone http://172.20.142.126:3000/evilliers/mitogenome-pipeline.git

# 2. Create a feature branch
git checkout -b fix/describe-your-change

# 3. Make your changes, then commit
git add run_pipeline.sh          # stage specific files, not git add -A
git commit -m "Fix: brief description of what changed"

# 4. Push the branch and open a PR on Gitea
git push origin fix/describe-your-change

Then open a pull request on Gitea at 172.20.142.126:3000/evilliers/mitogenome-pipeline/pulls. In the PR description, note:

  • What script(s) were changed and why
  • Which species / config was used to test
  • Any new or changed output files

Updating the SOP

If your change affects how the pipeline is run, update SOP_mitogenome_pipeline.md in the same PR. After merging, copy the updated SOP to the agrp-sops repository and push there too so the website stays in sync.

3. Prerequisites

3.1 Software and environments

The following tools must be installed before running the pipeline.

Tool Environment / method Used in step
NanoPlot ont conda env Step 1 (QC)
minimap2 ont conda env Step 3 (read enrichment)
samtools ont conda env Step 3 (read enrichment)
seqkit ont conda env Step 3, Step 6
Flye ont conda env Step 4 (assembly)
QUAST ont conda env Step 6 (stats)
medaka medaka conda env Step 5 (polishing)
MITOS2 (runmitos) mitos2 conda env Annotation script
BioPython mitos2 conda env Step 8 (BLAST species ID)
MitoZ Singularity image Step 7 (circos / gene order)
efetch (NCBI E-utilities) system PATH Step 0 (reference download)
Singularity / Apptainer system binary Step 7

The mitos2 conda environment and the MitoZ Singularity image are created automatically on first run if they do not exist. BioPython is already included in the mitos2 environment.

3.2 Input data requirements

  • ONT reads in .fastq.gz format (a single concatenated file per species)
  • A known NCBI RefSeq accession for a closely related mitogenome to use as a mapping reference

4. Repository layout

mitogenome-pipeline/
├── run_pipeline.sh          # Main assembly pipeline (Steps 0–8)
├── annotate_mitos2.sh       # MITOS2 annotation (run separately)
├── generate_report.sh       # Final report generator
├── configs/
│   └── anguilla_marmorata.conf   # Example species config
├── reference/               # Reference FASTAs (auto-downloaded, not in git)
│   └── refseq89m/           # MITOS2 reference database (auto-downloaded)
├── singularity/
│   └── mitoz_3.6.sif        # MitoZ Singularity image (auto-downloaded, not in git)
└── results/
    └── <species_prefix>/    # All outputs for a species (not in git)
        ├── 01_qc/
        ├── 03_mito_reads/
        ├── 04_assembly/
        ├── 05_polished/
        ├── 06_stats/
        ├── 07_mitoz/
        ├── 07_annotation/
        ├── 08_blast/
        └── report.md

Note: 07_annotation/ only appears after running annotate_mitos2.sh. 08_blast/ and report.md are created by run_pipeline.sh and generate_report.sh respectively.

5. Step 1 — Create a species config file

Each species requires a config file in the configs/ directory. Copy the example and edit it for the new species.

cp configs/anguilla_marmorata.conf configs/<new_species>.conf

Open the new file and set the following variables:

# =============================================================================
# Species config: <Common name or full name>
# =============================================================================

SPECIES_NAME="Genus species"         # Full scientific name (for display only)
SPECIES_PREFIX="genus_species"       # Lowercase, underscore-separated (used for file names)

# Input reads (fastq or fastq.gz) — path relative to project root
READS="path/to/reads.fastq.gz"

# NCBI RefSeq accession for a closely related mitogenome
REFERENCE_ACCESSION="NC_XXXXXX.1"

# Expected mitogenome size — used by Flye (e.g., 17k, 16.5k)
GENOME_SIZE="17k"

# Flye read type:
#   --nano-hq   for R10+ flowcells with high-accuracy basecalling (recommended)
#   --nano-raw  for older R9.4.1 flowcells or standard basecalling
FLYE_MODE="--nano-hq"

# Medaka polishing model — must match the flowcell and basecaller version used
# See: https://github.com/nanoporetech/medaka#models
MEDAKA_MODEL="r1041_e82_400bps_sup_v5.0.0"

Choosing REFERENCE_ACCESSION: Search NCBI for a complete mitogenome from the same genus or a closely related genus. The reference is used only for read enrichment (mapping) and QUAST comparison; it does not need to be from the same species.

Choosing FLYE_MODE:

Flowcell Basecalling Use
R10.4.1 + SUP/HAC --nano-hq
R9.4.1 --nano-raw

Choosing MEDAKA_MODEL: The model name encodes the flowcell chemistry, pore version, basecaller speed, and model version. Check medaka tools list_models for all available models.

6. Step 2 — Run the assembly and MitoZ pipeline

Activate the ont conda environment, then run the pipeline with the species config:

conda activate ont
bash run_pipeline.sh configs/<species>.conf

Example: 

conda activate ont
bash run_pipeline.sh configs/anguilla_marmorata.conf

The script will run non-interactively through all steps. The first run for a new species will also:

  • Download the reference mitogenome from NCBI (Step 0, requires internet)
  • Pull the MitoZ Singularity image (~1 GB, Step 7, requires internet)

Both are skipped on subsequent runs if the files already exist.

Pipeline steps

Step Tool Description Output directory
0 efetch Download reference mitogenome from NCBI reference/
1 NanoPlot Read quality report results/<prefix>/01_qc/
3 minimap2 + samtools Map all reads to reference, keep mapped reads results/<prefix>/03_mito_reads/
4 Flye Assemble enriched mito reads results/<prefix>/04_assembly/
5 Medaka Polish assembly with raw reads results/<prefix>/05_polished/
6 seqkit + QUAST Assembly statistics and comparison to reference results/<prefix>/06_stats/
7 MitoZ (Singularity) Circos plot and gene order results/<prefix>/07_mitoz/
8 BioPython + NCBI BLAST Species identification — top 10 nt hits results/<prefix>/08_blast/

Note

Step 2 (Filtlong pre-filtering) is intentionally skipped. Pre-filtering to a size limit before mapping discards most mitochondrial reads. All raw reads are mapped directly to the reference instead.

Note

Step 8 submits consensus.fasta to NCBI BLAST over HTTP and may take 5–15 minutes depending on NCBI load. An internet connection is required.

Checking assembly quality after Step 6

Before proceeding to annotation, verify the assembly looks correct:

  1. Check contig count and circularity — results/<prefix>/04_assembly/assembly_info.txt should show 1 contig, circ. = Y, length ~16–18 kb.
  2. Check QUAST genome fraction — results/<prefix>/06_stats/report.txt should show Genome fraction (%) = ~100 and # misassemblies = 0.
  3. Check MitoZ gene order — results/<prefix>/07_mitoz/<prefix>.consensus.fasta.result/ should contain the circos plot and GenBank file.

7. Step 3 — Run MITOS2 annotation

MITOS2 is run as a separate script. Pass the same config file used for assembly — the script derives all paths from SPECIES_PREFIX and writes output to results/<prefix>/07_annotation/.

bash annotate_mitos2.sh configs/<species>.conf

Example: 

bash annotate_mitos2.sh configs/anguilla_marmorata.conf

On first run, the script will automatically:

  1. Create the mitos2 conda environment (if not present)
  2. Download the MITOS2 reference database refseq89m from Zenodo (~20 MB)

MITOS2 output files

File Contents
result.bed Gene coordinates (BED format)
result.gff Gene annotation (GFF3 format)
result.fas Gene nucleotide sequences
result.faa Protein sequences
result.mito Summary of all annotated features
result.geneorder Gene order string
result.png Linear genome map

Expected result for a vertebrate fish: 13 protein-coding genes, 2 rRNAs (12S, 16S), 22 tRNAs. Missing or fragmented genes may indicate a misassembly or a genuinely unusual mitogenome and should be investigated.

8. Step 4 — Generate the final report

Once the pipeline and MITOS2 annotation are complete, generate a markdown report that summarises all results in one place:

bash generate_report.sh configs/<species>.conf

Example: 

bash generate_report.sh configs/anguilla_marmorata.conf

The report is written to results/<prefix>/report.md and contains:

  • Assembly statistics (contig length, coverage, circularity, QUAST metrics)
  • BLAST species identification table (top 10 NCBI nt hits)
  • MitoZ annotation summary and full gene table
  • MITOS2 annotation summary and gene order (if annotate_mitos2.sh has been run)
  • Index of all output files

The report can be opened in any markdown viewer, text editor, or converted to PDF/HTML. If MITOS2 has not yet been run, section 4 will display a prompt with the command to run it — the report can be regenerated at any time.

9. Interpreting outputs

Assembly quality indicators

Metric Acceptable range File
Contig length 15,500–18,000 bp results/<prefix>/04_assembly/assembly_info.txt
Circular Y results/<prefix>/04_assembly/assembly_info.txt
Coverage > 50× (ideally 100–200×) results/<prefix>/04_assembly/assembly_info.txt
Genome fraction > 95% results/<prefix>/06_stats/report.txt
Misassemblies 0 results/<prefix>/06_stats/report.txt
Mismatches per 100 kbp < 1,000 results/<prefix>/06_stats/report.txt

MitoZ circos plot

MitoZ names its result directory after the input filename, so the outputs are at:

results/<prefix>/07_mitoz/<prefix>.consensus.fasta.result/circos.png
results/<prefix>/07_mitoz/<prefix>.consensus.fasta.result/circos.svg

The GenBank annotation file is at:

results/<prefix>/07_mitoz/tmp_<prefix>_consensus.fasta_mitoscaf.fa/<prefix>_consensus.fasta_mitoscaf.fa.gbf

The circos plot shows the circular mitogenome with gene positions, strand orientation, and a GC content track. Both PNG (289 kb) and SVG (44 kb) are produced — open the SVG in any web browser or vector graphics editor for a scalable version.

BLAST species identification

Results are in results/<prefix>/08_blast/blast_results.tsv with columns: rank, accession, identity %, alignment length, e-value, bitscore, description.

The top hit identity percentage is the key value to interpret:

Identity % Interpretation
≥ 99% Strong species-level match
97–99% Likely same species, minor sequence divergence
95–97% Probable congeneric species
< 95% Genus-level or higher match only

If the top BLAST hits disagree with the sample label, treat the BLAST result as the authoritative identification. A mislabelled sample will show all top hits belonging to a different genus or family than expected.

Warning

MitoZ's "closely related species" in summary.txt is determined from its own limited internal database and is used to guide annotation — it is not a species identification. Always use the BLAST result for species ID.

Gene order

The gene order string from MITOS2 is in results/<prefix>/07_annotation/result.geneorder. A - prefix indicates the gene is on the minus strand. Compare against published gene orders for the taxon to confirm the assembly is correctly oriented.

10. Adding a new species — quick checklist

  • [ ] Obtain concatenated ONT reads (.fastq.gz)
  • [ ] Find a RefSeq accession for a related mitogenome on NCBI
  • [ ] Clone (or pull) the pipeline repo: git clone http://172.20.142.126:3000/evilliers/mitogenome-pipeline.git
  • [ ] Copy and edit a config: cp configs/anguilla_marmorata.conf configs/<new_species>.conf
  • [ ] Set all 7 config variables: SPECIES_NAME, SPECIES_PREFIX, READS, REFERENCE_ACCESSION, GENOME_SIZE, FLYE_MODE, MEDAKA_MODEL
  • [ ] Run: conda activate ont && bash run_pipeline.sh configs/<new_species>.conf
  • [ ] Check assembly quality (contig count, circularity, QUAST)
  • [ ] Verify BLAST top hit matches expected species/genus (results/<prefix>/08_blast/blast_results.tsv)
  • [ ] Run: bash annotate_mitos2.sh configs/<new_species>.conf
  • [ ] Verify gene count (expect 13 CDS + 2 rRNA + 22 tRNA)
  • [ ] Generate final report: bash generate_report.sh configs/<new_species>.conf
  • [ ] Commit the new config file and push (or open a PR): git add configs/<new_species>.conf && git commit -m "Add config: <species name>"

11. Troubleshooting

Very few reads after Step 3 (mito read enrichment) The reference may be too divergent from your species. Try a closer relative as the mapping reference. Check seqkit stats output after Step 3 — at minimum a few thousand reads are needed for assembly.

Flye produces 0 contigs or multiple short contigs Coverage is likely too low. Check the read count from Step 3. Consider using --nano-raw if reads were basecalled without super-accuracy mode.

QUAST shows low genome fraction or many misassemblies The reference may be from a distantly related species. This does not necessarily indicate a bad assembly — compare GC content and length against published mitogenomes for the taxon instead.

MitoZ Singularity pull fails Check internet connectivity and available disk space (df -h). The .sif file is ~1 GB. Re-run the pipeline once the image has been downloaded manually: 

singularity pull singularity/mitoz_3.6.sif docker://guanliangmeng/mitoz:3.6

BLAST step times out or returns no results NCBI BLAST can be slow under heavy load. The step may take up to 15 minutes. If it fails, the pipeline will exit — re-run from Step 8 by calling the BioPython command directly, or submit results/<prefix>/05_polished/consensus.fasta manually to NCBI BLAST and save the top 10 hits as results/<prefix>/08_blast/blast_results.tsv.

BLAST top hits do not match the sample label This is a strong indicator of sample mislabelling. Trust the BLAST result over the label. See section 9 (Interpreting BLAST results) for identity % thresholds. Note that MitoZ's "closely related species" is not a species identification — always use BLAST for this purpose.

MITOS2 misses genes Try the MITOS2 web server with the polished consensus as input (Reference: RefSeq 89 Metazoa, Genetic code: 2 — vertebrate mt). The web server uses the same algorithm but may give slightly different results with default settings.