bhargava-morampalli/rnamodbench

Introduction

rnamodbench is a Nextflow DSL2 pipeline for benchmarking RNA modification callers on Oxford Nanopore direct RNA sequencing data. It compares native RNA samples against in-vitro transcribed (IVT) controls using 9 modification detection tools.

Pipeline Summary

1. Validate inputs, parse samplesheet, and parse target-to-reference mapping.
2. Map each sample to its designated target reference and generate sorted/indexed BAMs.
3. Generate QC metrics (flagstat, depth, coverage plots, NanoPlot).
4. Prepare signal-level data (read IDs, FAST5 subset, multi-to-single FAST5 conversion).
5. Run signal processing (f5c index/eventalign and tombo resquiggle).
6. Run modification callers and aggregate software versions/reports.

Data Flow

graph TD
    A[samplesheet.csv] --> C[Input validation and metadata build]
    B[references.csv] --> C

    C --> D[Mapping: minimap2 -> samtools]
    D --> E[QC: flagstat, depth, coverage, nanoplot]

    D --> F[Extract mapped FASTQ]
    F --> G[Extract read IDs]
    G --> H[FAST5 subset]
    H --> I[Multi-to-single FAST5]

    D --> J[f5c index]
    H --> J
    J --> K[f5c eventalign read_name]
    J --> L[f5c eventalign_xpore read_index]

    I --> M[tombo resquiggle]

    M --> N[Tombo detect + text output]
    K --> O[Yanocomp prep + gmmtest]
    K --> P[Nanocompore collapse + sampcomp]
    L --> Q[xPore dataprep + diffmod]

    D --> R[BAM-based callers]
    R --> S[ELIGOS]
    R --> T[EpiNano-Error]
    R --> U[DiffErr]
    R --> V[DRUMMER]
    R --> W[JACUSA2]

    N --> X[modifications/]
    O --> X
    P --> X
    Q --> X
    S --> X
    T --> X
    U --> X
    V --> X
    W --> X

    X --> Y[pipeline_info/software_versions.yml]

Loading

Caller input routing:

• tombo: resquiggled single FAST5
• yanocomp, nanocompore: eventalign output with read names
• xpore: eventalign_xpore output with read index
• eligos, epinano, differr, drummer, jacusa2: paired native/IVT BAMs + target reference

Quick Start

1) Clone and run locally

git clone <your-repo-url> rnamodbench
cd rnamodbench

nextflow run main.nf \
  --input /absolute/path/samplesheet.csv \
  --references /absolute/path/references.csv \
  --outdir /absolute/path/results \
  -profile singularity

2) Resume an interrupted run

nextflow run main.nf \
  --input /absolute/path/samplesheet.csv \
  --references /absolute/path/references.csv \
  --outdir /absolute/path/results \
  -profile singularity \
  -resume

3) Common profiles

Profile	Description
singularity	Use Singularity containers
docker	Use Docker containers
conda / mamba	Use Conda/Mamba environments
test	Run with minimal test data

Inputs

Samplesheet (--input)

Required CSV columns:

Note: sample CSVs in this repository use placeholder paths and must be edited for your local filesystems and datasets.

Column	Description
sample	Unique sample ID (no spaces)
fastq	FASTQ path (.fq, .fastq, optionally .gz)
type	native or ivt
replicate	Replicate identifier (rep1, rep2, ...)
fast5_dir	Directory containing FAST5 files
target	Target label (for example 16s, 23s, 5s)

Example:

sample,fastq,type,replicate,fast5_dir,target
native_16s_rep1,/data/native_16s_rep1.fastq.gz,native,rep1,/data/native_fast5,16s
ivt_16s_rep1,/data/ivt_16s_rep1.fastq.gz,ivt,rep1,/data/ivt_fast5,16s
native_23s_rep1,/data/native_23s_rep1.fastq.gz,native,rep1,/data/native_fast5,23s
ivt_23s_rep1,/data/ivt_23s_rep1.fastq.gz,ivt,rep1,/data/ivt_fast5,23s

Reference map (--references)

Required CSV columns:

Column	Description
target	Target label matching samplesheet target
reference	FASTA path for that target

Example:

target,reference
16s,/refs/k12_16S.fa
23s,/refs/k12_23S.fa

Modification Callers

Tool	Approach	Input
Tombo	Signal-level de novo detection	Resquiggled FAST5
Yanocomp	GMM-based current comparison	Eventalign HDF5
Nanocompore	Sample comparison	Eventalign collapsed
xPore	Bayesian differential modification	Eventalign (read index)
ELIGOS	Error signature analysis	Paired BAMs
EpiNano-Error	Mismatch-based comparison	Paired BAMs
DiffErr	Differential error rate	Paired BAMs
DRUMMER	Odds-ratio based detection	Paired BAMs
JACUSA2	Variant-like site detection	Paired BAMs

Each target + replicate requires both native and ivt samples. See docs/usage.md for parameter details.

Output Structure

Key output directories under --outdir:

Directory	Contents
mapping/	Sorted BAMs and indices
qc/	Flagstat, depth, coverage, NanoPlot
modifications/	Per-tool modification calls
logs/	Per-tool log files
pipeline_info/	Execution reports, timeline, DAG, software versions
downstream_analysis/	Benchmark metrics, visualizations (if --run_downstream true)

See docs/output.md for full details.

Documentation

Document	Description
docs/usage.md	Parameters and configuration
docs/output.md	Output file descriptions
docs/downstream_analysis.md	Downstream analysis usage
CITATIONS.md	Tool citations

Credits

This pipeline was developed by Bhargava Morampalli.

Support

For issues and feature requests, open an issue in this repository.