ECHO: a nanopore sequencing-based workflow for (epi)genetic profiling of the human repeatome

The paper introduces ECHO, a user-friendly, Snakemake-based pipeline that leverages Oxford Nanopore sequencing to enable comprehensive, reproducible, and scalable end-to-end (epi)genetic profiling of the human repeatome.

The Gist

Imagine the human genome as a massive, ancient library. For a long time, scientists were only interested in the "bestsellers"—the unique, readable books that tell us how to build our bodies. They largely ignored the rest of the library because it was filled with repetitive noise: thousands of copies of the same sentence, pages stuck together, and books that looked exactly like each other.

For years, this "repetitive section" (called the repeatome) was considered junk. But we now know it's actually a control room. It holds the switches that turn genes on and off, and when it breaks, it can cause diseases like Alzheimer's or cancer.

The problem? The old tools scientists used were like short-sighted librarians. They could only read a few words at a time. If they tried to read a page full of repeated sentences, they got lost and couldn't tell one copy from another.

Enter ECHO: The "Super-Librarian"

This paper introduces a new tool called ECHO (which stands for Epigenomic Characterisation of Human O... well, let's just call it the Echo because it listens to the whole library at once).

ECHO is a software pipeline designed to work with a special kind of high-tech scanner (Oxford Nanopore sequencing) that can read entire long books in one go, rather than just a few words.

Here is how ECHO works, using some everyday analogies:

1. The "Noise-Canceling" Headset (Preprocessing)

Before reading the books, ECHO puts on noise-canceling headphones. It takes the raw, messy data from the scanner and cleans it up. It filters out the static, organizes the pages, and makes sure the text is clear.

2. The "Twin Detective" (Phasing)

This is ECHO's superpower. Humans have two copies of every book in their library (one from mom, one from dad). Often, these copies have tiny differences.

• Old tools would mix the two copies together, like shuffling two decks of cards and trying to read them as one pile.
• ECHO acts like a twin detective. It separates the "Mom deck" from the "Dad deck." It can tell you exactly which version of a repetitive sentence belongs to which parent. This is crucial because sometimes the "Mom" version is broken, but the "Dad" version is fine.

3. The "Two-Part Report" (Profiling)

Once the books are separated, ECHO generates a detailed report on two things:

• The Text (Genetics): It counts how many times a sentence is repeated. Is it 5 times? 50 times? 500 times? If a sentence is repeated too many times, it might be a sign of a disease.
• The Highlighter (Epigenetics): This is the magic part. Imagine someone used a highlighter to mark certain pages. In our DNA, this "highlighter" is methylation. It doesn't change the words, but it tells the cell, "Don't read this part," or "Read this part loudly." ECHO can see exactly where these highlights are, even on the messy, repetitive pages that other tools miss.

Why is this a big deal?

Think of the human genome as a giant, tangled ball of yarn.

• Old methods tried to untangle it by cutting small pieces. They could only see the knots on the outside, missing the complex tangles deep inside.
• ECHO is like a robot that can gently pull the whole ball apart, thread by thread, without breaking it. It can see the tangles, count the knots, and see where the dye (methylation) was applied.

The Results

The authors tested ECHO on a "gold standard" human genome (HG002). They found that:

1. It's accurate: Its reading of the "highlighters" (methylation) matched the gold-standard lab tests almost perfectly (95-96% match).
2. It's fast: It can process a whole human genome in about a day and a half on a standard computer cluster.
3. It's open: It's free for anyone to use, like a shared toolkit for scientists.

The Bottom Line

ECHO is a new, user-friendly "Swiss Army Knife" for scientists. It allows them to finally explore the "junk" section of our DNA library with clarity. By understanding these repetitive regions and their chemical switches, we can better understand how our bodies work, why diseases happen, and potentially find new ways to treat them.

In short: ECHO helps us finally read the parts of our genetic instruction manual that we were previously too blind to see.

Technical Summary

1. Problem Statement

The human genome is dominated by repetitive DNA elements (over 50%), including Tandem Repeats (TRs) and Transposable Elements (TEs). Historically, these regions have been underexplored due to the limitations of short-read sequencing technologies, which struggle to resolve long, complex, and repetitive sequences. While recent Long-Read Sequencing (LRS) technologies, specifically Oxford Nanopore Technologies (ONT), offer the ability to sequence through these regions and directly detect DNA methylation, current bioinformatics tools remain fragmented.

• The Gap: Existing tools typically focus on specific repeat classes (either TRs or TEs) or specific sequencing platforms. There is a lack of a unified, comprehensive pipeline that can simultaneously characterize the sequence variation and epigenetic state (DNA methylation) of diverse repeat types across the whole genome in a haplotype-resolved manner.

2. Methodology: The ECHO Pipeline

The authors present ECHO (Epi)genomic Characterisation of Human Repetitive Elements using Oxford Nanopore Sequencing), a user-friendly, reproducible, and scalable workflow built on Snakemake. The pipeline is designed to run on Unix-based High-Performance Computing (HPC) systems or local servers, utilizing Singularity containers for dependency management.

The workflow consists of two core modules:

Module I: Preprocessing and Phasing

• Input: Accepts raw ONT data (POD5) or basecalled files (UBAM, FASTQ, BAM) generated with methylation-aware models.
• Basecalling & QC: Uses Dorado (SUP mode) for basecalling if raw data is provided. Performs quality control and optional read filtering using Chopper.
• Alignment: Maps reads to reference genomes (GRCh38 or T2T-CHM13v2) using minimap2.
• Variant Calling: Identifies SNVs/INDELs using Clair3 and Structural Variants (SVs) using Sniffles2.
• Phasing: Utilizes LongPhase to co-phase small variants and SVs using methylation information, generating haplotype-resolved BAM and VCF files.

Module II: Repeatome Profiling

This module performs integrative genotyping and methylation analysis for TRs and TEs:

• Genome-wide Methylation: Uses modkit to generate phased and unphased methylation pileups.
• Tandem Repeats (TRs):
  • Employs LongTR for genotyping a user-defined catalogue of TR loci.
  • Extracts overlapping reads in a haplotype-aware manner and performs local realignment.
  • Quantifies methylation at single-CpG resolution and regional averages (within the repeat and flanking regions) using modkit.
  • Uses uTR to decompose motif composition.
• Transposable Elements (TEs):
  • Reference TEs (ref-TEs): Intersects phased variants with RepeatMasker annotations to quantify sequence variation and calculates allele-specific methylation.
  • Non-Reference TEs (non-ref-TEs): Uses TLDR to detect novel TE insertions. ECHO then processes TLDR outputs to calculate methylation profiles for these new insertions and their flanking regions.

3. Key Contributions

• Unified Framework: ECHO is the first pipeline to integrate the analysis of both TRs and TEs (including non-reference insertions) alongside DNA methylation profiling in a single, end-to-end workflow.
• Haplotype Resolution: Unlike many existing tools, ECHO provides haplotype-resolved (phased) genotyping and methylation data, allowing researchers to distinguish epigenetic patterns between maternal and paternal alleles.
• Scalability and Portability: Built with Snakemake and Singularity, the pipeline is designed for reproducibility and can be deployed on various HPC environments.
• Customizability: Supports both genome-wide analysis and targeted panels via user-defined BED catalogues.

4. Results and Performance Evaluation

The pipeline was benchmarked using publicly available HG002 (Genome in a Bottle) ONT data at 30× and 15× coverage, compared against Whole-Genome Bisulfite Sequencing (WGBS) as the gold standard.

• Repeat Detection:
  • At 30× coverage, the pipeline successfully genotyped a vast number of TR loci (using the LongTR catalogue) and detected non-reference TE insertions (using TLDR).
  • A moderate reduction in detected loci was observed at 15× coverage, but concordance remained high.
• Methylation Accuracy:
  • ECHO showed high concordance with WGBS data.
  • Pearson Correlation Coefficients (r):
    • Genome-wide: 0.96
    • Within TEs: 0.95
    • Within TRs: 0.94
  • These results demonstrate that ONT-based methylation calling within complex repetitive regions is as accurate as genome-wide measurements.
• Efficiency:
  • Runtime: 38.5 hours (234 CPU-hours) for the 30× dataset.
  • Storage: Final project directories occupied ~100 GB (30×) and ~60 GB (15×).

5. Significance

ECHO addresses a critical bottleneck in human genomics by enabling the comprehensive, simultaneous characterization of the "human repeatome." By bridging the gap between sequence variation and epigenetic regulation in repetitive regions, ECHO facilitates:

• Disease Research: Better understanding of repeat expansion disorders (neurological/developmental) and TE-mediated tumorigenesis.
• Regulatory Insights: Uncovering how dynamic, context-specific DNA methylation patterns in repeats influence gene regulation and genome stability across different cell types and disease states.
• Future-Proofing: The modular design allows for the easy integration of future tools and updates, ensuring long-term utility in the rapidly evolving field of long-read sequencing.

The tool is freely available on GitHub and Zenodo, promoting open science and reproducibility in repeatome research.