ProkaryONT 🧬

ProkaryONT is a comprehensive, production-grade automated pipeline for Oxford Nanopore Technologies (ONT)-based prokaryotic genome assembly, polishing, taxonomic classification, quality assessment, and high-fidelity ensemble functional annotation.

It orchestrates industry-standard bioinformatics tools via a robust, subcommand-driven shell interface, allowing both rapid automated runs and hands-on, step-by-step manual curation.

---

🗺️ Pipeline Overview

The pipeline is split into three main execution phases: Core Assembly & Polishing (Stages 1–4), Primary Downstream Analysis (Taxonomy & Single-Tool Annotation), and an Ensemble Annotation & Reconcilation Pipeline.

graph TD
    A[Raw ONT FASTQ + Summary] --> Stage1[Stage 1: QC & Estimate<br/>NanoPlot / LRGE / Raven / Meryl<br/>+ Fastcat / Porechop_ABI scan / SNIKT]
    Stage1 --> Stage2[Stage 2: Preprocess & Filter<br/>Porechop_ABI / Chopper / Seqkit / Filtlong<br/>+ Fastplong QC report]
    Stage2 --> Stage3[Stage 3: Autocycler Assembly]
    Stage3 --> C1{"🔍 Curation 1: Assemblies"}
    C1 --> Stage2b[Clustering & tree QC<br/>autocycler cluster / ETE3]
    Stage2b --> C2{"🔍 Curation 2: Clusters"}
    C2 --> Stage2c[Trim, Resolve & Combine<br/>autocycler trim/resolve/combine / MUMmer]
    Stage2c --> C3{"🔍 Curation 3: Dotplots"}
    C3 --> Stage2d[Consensus Assembly + Depth Mapping]
    Stage2d --> Stage4[Stage 4: Polish & Orient<br/>Dorado Polishing / Dnaapler Reorientation]
    Stage4 --> FinalAssembly[Final Assembly: dnaapler_reoriented.fasta]
    
    FinalAssembly --> TaxStage[Taxonomy Stage<br/>GTDB-Tk / mlst / Barrnap 16S]
    FinalAssembly --> AssessStage[Assess & Annotate Stage<br/>Bakta / QUAST / CheckM2 / BUSCO / Merqury]
    
    FinalAssembly --> EnsembleStage[Ensemble Annotation Pipeline]
    EnsembleStage --> 07a[07a: Gene Prediction<br/>Pyrodigal / Glimmer3 / GeneMarkS-2]
    EnsembleStage --> 07b[07b: Track A Annotation<br/>Bakta / DRAM / RASTtk]
    EnsembleStage --> 07c[07c: Track B Annotation<br/>eggNOG-mapper / InterProScan / KofamScan / DIAMOND SwissProt]
    EnsembleStage --> 07d[07d: Track C Annotation<br/>geNomad / Phanotate]
    07a & 07b & 07c & 07d --> 07e[07e: Reconcile & Merge<br/>GFF3 / TSV / GenBank / JSON]

Loading

---

✨ Key Features

• Automated Genome Size Estimation: Integrates weighted estimates from LRGE (k-mer-based) and a fast Raven draft assembly ((2*Raven + LRGE) / 3) to configure downstream subsampling.
• Robust Subsample Assemblies: Subsamples input reads and runs up to 10 assemblers in parallel (including Flye, Canu, Hifiasm, Raven, Miniasm, MetaMDBG, Myloasm, Plassembler, NextDenovo, and Wtdbg2) using GNU Parallel.
• ETE3-Based Tree Checks: Evaluates the clustering tree dynamically to flag clades showing branch lengths exceeding $5\times$ the median or single-assembler dominance.
• Automated Plot Generation: Distributes pairwise alignments using Autocycler dotplot for small replicons (plasmids) and MUMmer (nucmer + mummerplot) for chromosome-scale sequences to review overlap trimming.
• Real-Time Read-Depth Tracking: Maps filtered reads back to consensus using minimap2 and samtools to generate coverage metrics and flag high-copy-number replicons (depth $> 1.5\times$ relative to the chromosome).
• Dorado-Based Polishing: Checks for moves tables (mv:B:c tags) in BAM files and prompts for read group resolution (unification, selection, or ignoring) in multi-read-group runs.
• Smart Reorientation: Utilizes Dnaapler to locate start genes (e.g., dnaA for chromosomes, repA for plasmids, terL for phages) and rotate contigs to a standardized biological origin.
• Ensemble Annotation Engine:
  • Genomic Predictions: Merges predictions from Pyrodigal, Glimmer3, and GeneMarkS-2 using a consensus algorithm with a Pyrodigal tiebreaker.
  • Assembly Features (Track A): Coordinates structural annotation via Bakta, DRAM, and imported BV-BRC RASTtk records.
  • Homology Databases (Track B): Consolidates predictions from eggNOG-mapper, InterProScan, KofamScan, and DIAMOND SwissProt.
  • Prophage Discovery (Track C): Extracts prophage intervals using geNomad and annotates phage-specific ORFs with Phanotate.
  • Consensus Reconciliation: Merges all annotation data into a single GFF3, TSV annotation matrix, fully annotated GenBank flatfile, and an isolate summary JSON.

---

🛠️ Installation & Dependencies

To run the pipeline, ensure the following tools are available in your $PATH:

Phase 1: Assembly & Polishing

• Filtering & QC: filtlong, seqkit, NanoPlot, fastcat, porechop_abi, snikt.R, chopper, fastplong
• Draft Estimation: lrge, raven
• Subsampling & Assembly: autocycler (v0.1.0+), meryl, samtools, GNU parallel, and desired assemblers (flye, canu, hifiasm, metaMDBG, miniasm, minipolish, minimap2, plassembler, etc.)
• Tree Analysis: Python 3 with ete3 library (optional)
• Dotplots: mummer (specifically nucmer, mummerplot)
• Polishing & Reorientation: dorado, dnaapler

Phase 2: Downstream & Annotation

• Taxonomy: gtdbtk (requires GTDB reference DB), mlst, barrnap
• Assessment: quast, checkm2, busco (with lineages), merqury (requires merqury.sh)
• Gene Finding: pyrodigal, glimmer3
• Ensemble Annotation: bakta (requires database), DRAM.py, emapper.py (eggNOG-mapper), interproscan.sh, exec_annotation (KofamScan), diamond, genomad (requires database), phanotate.py

---

⚙️ Configuration (pipeline.conf)

The configuration file pipeline.conf controls all paths, database locations, threads, and parameter thresholds. Key variables include:

# --- Required inputs ---
input_fastq=input.fastq.gz
sequencing_summary=sequencing_summary.txt
pod5_dir=pod5_dir/
sample_name=MyBacteria

# --- Compute & Filtering ---
threads=128
read_type=ont_r10
filtlong_min_length=200
filtlong_keep_percent=90

# --- Databases ---
gtdbtk_data_path=/path/to/gtdbtk_db
bakta_db=/path/to/bakta_db
genomad_db=/path/to/genomad_db
interproscan_db=/path/to/interproscan
kofam_profiles=/path/to/kofam/profiles
kofam_ko_list=/path/to/kofam/ko_list

---

🚀 Execution & Usage

The main interface to the pipeline is prokaryont.sh.

chmod +x prokaryont.sh
./prokaryont.sh --help

1. Full Automated Assembly (Stages 1–4)

Runs read QC, filtering, multi-assembler subsampling, clustering, consensus combination, dorado polishing, and dnaapler reorientation.

./prokaryont.sh assemble --config pipeline.conf

Note

During assembly, the pipeline will pause at Manual Curation Points to allow file inspection (e.g., verifying draft assembly sizes, moving misclustered contigs, inspecting dotplots). You can bypass these pauses by passing the --skip-curation flag (or uncommenting skip_curation=true in the configuration).

2. Resuming Execution

If the pipeline halts during assembly or you wish to adjust parameters, you can resume from specific steps using:

./prokaryont.sh assemble --config pipeline.conf --resume-from [cluster|trim|dnaapler]

• cluster: Skips read filtering and assemblies, resumes from Autocycler clustering. (post-split: resumes Stage 3 from clustering, skipping the subsample/multi-assembler step — assumes filtered_input.fastq.gz from Stage 2 already exists)
• trim: Skips assembly and clustering, resumes from Autocycler overlap trimming/resolving.
• dnaapler: Skips polishing, resumes from Dnaapler reorientation.

3. Taxonomic Classification

Executes GTDB-Tk classification, MLST typing, and Barrnap 16S rRNA gene extraction.

./prokaryont.sh taxonomy --config pipeline.conf

Review the taxonomy outputs to update your organism details (genus, species, gram stain) in pipeline.conf prior to annotation.

4. Primary Annotation & QA

Runs Bakta annotation and standard quality assessment metrics (QUAST, CheckM2, BUSCO, and Merqury).

./prokaryont.sh annotate --config pipeline.conf

5. Ensemble Annotation Pipeline

For high-fidelity structural and functional annotation, run the ensemble scripts sequentially:

# A. Predict consensus gene coordinates
bash 07a_predict_genes.sh --assembly dnaapler_reoriented.fasta --config pipeline.conf

# B. Run assembly-level annotations (Track A)
bash 07b_annotate_trackA.sh --assembly dnaapler_reoriented.fasta --bakta-db /path/to/bakta_db

# C. Run protein-level sequential annotations (Track B)
bash 07c_annotate_trackB.sh --consensus-proteins 13_gene_prediction/consensus/consensus_proteins.faa --config pipeline.conf

# D. Run prophage annotation (Track C)
bash 07d_annotate_trackC.sh --assembly dnaapler_reoriented.fasta --genomad-db /path/to/genomad_db

# E. Reconcile and merge all annotation files
bash 07e_reconcile_merge.sh \
    --assembly dnaapler_reoriented.fasta \
    --consensus-gff 13_gene_prediction/consensus/consensus_genes.gff3 \
    --bakta-dir 14_trackA/bakta \
    --emapper-out 15_trackB/eggnog \
    --interproscan-out 15_trackB/interproscan \
    --kofamscan-out 15_trackB/kofamscan \
    --phanotate-gff 16_trackC_prophage/phanotate/phanotate_genes.gff3 \
    --sample-name MyBacteria

---

📁 Output Directory Directory Structure

Following a full run, the workspace directory will contain:

├── 01_qc/                        # NanoPlot quality reports for raw/filtered reads
├── 02_genome_size/               # LRGE, Raven estimates, and Meryl histograms
├── assemblies/                   # Fasta and job logs from subsampled assemblers
├── autocycler_out/               # Autocycler compressed outputs and GFA clusters
│   ├── cluster_metadata.tsv      # Per-cluster assembly and resolve metrics
│   └── consensus_assembly.fasta  # Combined draft assembly
├── 07_taxonomy/                  # GTDB-Tk, MLST, and Barrnap 16S fasta
├── bakta_result/                 # Single-run Bakta annotation
├── 10_quast/                     # QUAST assembly statistics
├── 11_checkm2/                   # CheckM2 quality reports (completeness/contamination)
├── 12_busco/                     # BUSCO conserved single-copy ortholog checks
├── 13_gene_prediction/           # Multi-caller prediction outputs (Pyrodigal/Glimmer)
├── 14_trackA/                    # Track A annotators (Bakta, DRAM)
├── 15_trackB/                    # Track B database searches (eggNOG, Pfam, Kofam)
├── 16_trackC_prophage/           # geNomad & Phanotate mobile element annotations
├── 17_final_annotation/          # Final integrated ensemble annotations
│   ├── final_annotation.gff3     # Fully reconciled GFF3 file
│   ├── final_annotation.gbk      # Fully reconciled GenBank file
│   ├── annotation_matrix.tsv     # Gene-by-annotation matrix
│   └── summary.json              # Isolate summary statistics
├── polished_assembly.fasta       # Dorado-polished assembly
├── dnaapler_reoriented.fasta     # Polished and reoriented final genome
├── contig_characteristics.tsv    # Combined length, cluster, depth, and copy number flags
├── metrics.tsv                   # Autocycler metrics table with read statistics
└── pipeline.log                  # Global execution log of the pipeline runs

---

📝 License

This project is licensed under the Apache License 2.0 - see the LICENSE file for details.