elarodon 0.1.1

A comprehensive pipeline for detection and analysis of large genomic rearrangements from Oxford Nanopore Technologies (ONT) sequencing data.

eLaRodON

A comprehensive pipeline for detection and analysis of large genomic rearrangements from Oxford Nanopore Technologies (ONT) sequencing data.

Table of Contents

• Overview
• Installation
• Quick Start
• Detailed Usage
• Output Files
• Example
• Troubleshooting
• Citation
• License

Overview

The ONTLRcaller pipeline performs detection and characterization of large genomic rearrangements through four integrated modules:

Detailed description of the algorithm

The eLaRodON algorithm is a specialized computational pipeline designed for comprehensive detection of large genomic rearrangements (LGRs) from Oxford Nanopore sequencing data. Unlike conventional tools developed primarily for germline variants, eLaRodON incorporates several innovative features specifically optimized for identifying somatic LGRs, including those supported by single reads. Input Processing

The algorithm begins by processing aligned sequencing data in BAM format. It performs chromosome-by-chromosome analysis to optimize memory usage, with an option to focus on specific genomic regions of interest. The tool only considers primary alignments containing complete mapping information to ensure analysis quality.

Core Detection Mechanism

1. Split-read Analysis:

• Identifies reads with fragments mapped to different genomic locations
• Detects strand changes and structural variants >50bp through CIGAR tag analysis
• Extracts all insertions and deletions from primary alignments

2. Junction Characterization:

Records all junction sites with their genomic features in two separate files:

• fusions.csv: Contains genome region junctions
• insertions.csv: Stores CIGAR-derived insertions

For fusion events, combines split reads corresponding to:

• Translocation boundaries (TRL)
• Inversion breakpoints (INV)
• Tandem duplication junctions (TD)

Variant Classification

The algorithm employs a sophisticated classification system that:

Merges Similar Events:

• Combines fusions and insertions across genomic regions
• Uses precise coordinates, strand orientation, and junction characteristics
• Optional merging via companion script for mechanistic studies

Insertion Sequence Analysis:

• Maps CIGAR-derived insertion sequences to reference genome using minimap2
• Reclassifies as tandem duplications when sequences map near original positions

Structural Annotation:

• Determines LGR types using strand orientation and junction characteristics

• Evaluates four key genomic features for each rearrangement:

  • Proximity to repeat sequences and mobile elements (via vcfanno)
  • Presence of 2-5 nucleotide microhomology
  • ≥80% sequence similarity over ≤35 nucleotide homeology between breakpoints
  • Presence of novel inserted sequences not matching either breakpoint region

Quality Assessment

eLaRodON incorporates several quality control metrics:

New Sequence Pattern (NSP) Scoring:

score=∑x ∈[An, Tn,Gn,Cn]1L(x)−L(s)/100score=∑x ∈[An​, Tn​,Gn​,Cn​]​1L(x)​−L(s)/100

- Identifies homopolymeric tracts (≥4bp) in novel junction sequences
- Helps discriminate true rearrangements from artifacts

Output Generation

The final output includes:

1. Comprehensive VCF File:

• Special tags for LGRs sharing coordinates but differing in variant type
• Annotations for:
  • Intersection sequences (ISMFS tag)
  • Novel inserted sequences (NSMFS tag)
  • Microhomology/homeology patterns
  • Repeat element proximity

2. Additional Outputs:

• Raw junction calls for downstream analysis
• Quality metrics for each detected variant
• Intermediate files for debugging and method development

Technical Innovations

Key algorithmic advancements include:

• Treatment of multiple split-reads as single sequences for accurate two-boundary detection
• Optional non-merging of similar LGRs to preserve mechanistic signatures
• Memory-efficient chromosomal processing
• Specialized handling of Nanopore-specific artifacts

The tool demonstrates particular strength in identifying complex rearrangements that most existing algorithms miss, including multiple tandem duplications, non-reciprocal translocations, and inversions with two defined boundaries. Its performance has been validated across multiple datasets, showing superior accuracy compared to existing tools like Sniffles2, NanoSV, and SVIM, particularly for variants in repetitive regions and those with low read support.

Installation

Prerequisites:

Python 3.7+

minimap2 (v2.24+)

vcfanno (v0.3.3+)

samtools (v1.15+)

htslib (v1.21+)

git (for cloning)

💡 Pro tip: see full list of program versions in requirements.txt

1. Clone the Repository

git clone https://github.com/aakechin/eLaRodON.git

2. Install Dependencies

pip install -r requirements.txt

3. Verify Installation

python3 ./eLaRodON-main/main.py -h

No errors = successful installation

Quick Start

Basic Pipeline Execution

python3 main.py \
    -dir ./results_elarodon \
    -bam sample.bam \
    -ref hg38.fa \
    -vcfanno path/to/vcfanno \
    -bed annotations.bed \
    -th 4

Detailed Usage

Core Arguments

Parameter	Required	Description
-bam, --bam_file	Yes	BAM file
-dir, --workdir	Yes	Output directory
-ref, --ref-genome	Yes	Reference genome FASTA
-bed, --bed-file	Yes	Annotation BED file
-vcfanno, --vcf-anno	Yes	vcfanno executable path

Processing Parameters

Parameter	Default	Required	Description
-div, --divide-chroms	False	No	Divide chromosome to analyze
-dvlen, --div-length	None	No	Length of regions for division chromosome to analyze
-len, --minimal-length	50	No	Min variant length (bp)
-clip, --minimal-clipped-length	100	No	Min clipped length (bp)
-dist,-dist-to-join-trl	1000	No	Min clipped length (bp)
-join,-maximal-distance-join	30	No	Max distance for fusion joining (bp)
-th, --threads	4	No	CPU threads
-cont, --continue	all	No	Name of stage for start: bam, join, def

Special file names

Parameter	Default	Required	Description
-in, --input-files	auto	No	Regular expression for CSV files
-lrs, --output-lrs	auto	No	CSV output file for LGRs
-ins, --output-ins	auto	No	CSV output file for INS
-sam, --sam_file	auto	No	SAM file with INS alignment

Output Control

Parameter	Default	Description
-out, --out-vcf	auto	VCF output filename
-nrt_ins, --not-remove-trash-align	False	Keep temp alignment files
-nrt_anno, --not-remove-trash-anno	False	Keep temp annotation files

To view all parameters and their descriptions, you can use

python3 main.py -h

Output Files

Main Outputs

*.junction_stat.LRs_join100.csv - Merged large rearrangements

*.junction_stat.INS_join100.csv - Insertion calls

*_all_LGRS.vcf - Final annotated variants

Example

Input Preparation

samtools index sample.bam

Full Analysis

python3 main.py \
    -bam sample.bam \
    -dir lr_results \
    -ref hg38.fa \
    -vcfanno ~/tools/vcfanno \
    -bed repeats.bed \
    -th 8 \

Expected Output vcf

#CHROM	POS	ID	REF	ALT	QUAL	FILTER	INFO	FORMAT	SAMPLE_NAME
chr12	3456789	LR1	N	DEL	0.78	PASS	SVTYPE=DEL;SVLEN=1200;CHROM2=chr12...	GT:DP:AD:VF	0/1:330:1,329:0.0
chr13	4123456	INS1	N	INS	0.65	PASS	SVTYPE=INS;SVLEN=350...	GT:DP:AD:VF	0/1:244:1,243:0.0

Troubleshooting

Common Issues

Missing dependencies:

Error: minimap2 not found in PATH

Solution: Install via bioconda or add to PATH

Memory errors:

Killed (process exited)

Solution: Reduce thread count or increase memory

Too many open files

Solution:

Run before eLaRodN:

ulimit -n 4096

BAM index missing:

    [E::idx_find_and_load] Could not retrieve index file for 'sample.bam'

Solution: Run samtools index sample.bam

Citation

Please cite:

eLaRodON: identification of large genomic rearrangements in Oxford Nanopore sequencing data

License

MIT License