Direct RNA Sequencing Analysis: From Raw Signals to Differential Methylation

At a glance:

• Introduction
• Step 1. Processing Raw Nanopore Signal Files
• Step 2. Visualizing Signals with Squigualiser
• Step 3. Identifying RNA Modifications
• Step 4. Differential Methylation Analysis
• Putting It Together: A Reproducible DRS Playbook
• Case-Oriented Scenarios You Can Reproduce
• Common Pitfalls and How to Avoid Them
• Tooling Notes and Ecosystem Bridges
• Future Directions and Research Opportunities
• Conclusion

Introduction

Direct RNA Sequencing analysis delivers native, molecule-level insight that conventional RNA-seq misses. It reads RNA directly, avoiding cDNA conversion and PCR amplification steps. That design preserves full-length structure, poly(A) tails, and chemical marks on RNA. It also reduces biases that appear during reverse transcription and amplification. The result is a more faithful view of transcript diversity and regulation.

This pillar post walks through a practical, end-to-end workflow. We start with raw signal management. We then cover signal visualization, modification calling, and differential methylation analysis. Each section includes actionable steps, tool options, and interpretation tips. You will also find planned URLs for four deep-dive posts you can publish later. Those posts will strengthen internal linking and topic authority for your site.

Target keywords appear naturally in this introduction. These include direct RNA sequencing analysis, RNA methylation detection, and nanopore raw signal processing. The goal is clarity for researchers and strong coverage for important search intents.

Nanopore technology can directly sequence intact RNA molecules

Step 1. Processing Raw Nanopore Signal Files

Nanopore instruments write electrical signal data as streaming binary files. The current standard is the POD5 format. POD5 supports efficient streaming writes from the sequencer and fast reads by analysis tools. It uses Apache Arrow under the hood, which enables performant cross-language access. These choices speed interactive QC and downstream processing at lab scale. The official POD5 toolkit bundles several utilities that simplify routine tasks:

• pod5 view builds a tabular summary for quick inspection. You can include only fields you need.
• pod5 inspect reports file-wide integrity metrics and per-read detail on demand.
• pod5 merge joins many POD5 files into a single container for batch analysis.
• pod5 filter extracts a read subset from a large file using an ID list.
• pod5 subset separates reads by barcode or QC status into structured outputs.
• pod5 repack re-packages files to improve compute performance during analysis.
• pod5 convert converts between POD5 and the legacy FAST5 format when required.

Why this step matters

• Clean, indexed inputs reduce friction throughout the pipeline.
• Repacking improves I/O and makes GPU basecalling more predictable.
• Merging and subsetting simplify sample management in large projects.

Practical tips

• Standardize directory layouts before large runs. It improves collaboration and reuse.
• Version your POD5 toolkit along with your pipeline.
• Document which fields you export with pod5 view for reproducibility.

If you want to explore this stage in greater detail—including practical examples of using pod5 view, pod5 inspect, and efficient strategies for merging, repacking, and visualizing nanopore signals—you can check out our comprehensive guide on Direct RNA Sequencing signal handling and visualization. This article expands on workflow tips and best practices that ensure high-quality inputs and interpretable outputs for downstream analyses.

Step 2. Visualizing Signals with Squigualiser

Raw current traces contain information about bases and modifications. Visualization bridges human intuition and machine decoding. Squigualiser aligns signal to sequence and renders interactive, single-base resolution plots. It supports both signal-to-read and signal-to-reference contexts. Multiple reads can be displayed as a pileup over a region. These features help you review basecalls, motifs, and putative modification signatures.

Typical workflow

• Basecall with Dorado and enable --emit-moves to output signal segmentation.
• Run squigualiser reform to restructure the BAM for plotting.
• Use samtools fasta to extract sequences for alignment.
• Align against a genome or transcriptome with minimap2.
• Convert POD5 to BLOW5 using blue-crab or the SLOW5 ecosystem.
• Plot with squigualiser plot and explore interactive HTML outputs.

Interpretation cues

• X-axis marks bases; colors distinguish nucleotides across aligned reads.
• Y-axis represents current intensity; deviations may indicate modifications.
• Signal pileups reveal consistent patterns and outliers across reads.

When to visualize

• New flowcell chemistries, models, or sample types.
• Suspected motif-linked artifacts or basecalling drift.
• Educational walkthroughs for new team members.

Step 3. Identifying RNA Modifications

Modification calling is a signature advantage of direct RNA sequencing. Dorado can detect several base modifications during basecalling. Supported calls include m6A, m5C, pseudouridine (Ψ), and A-to-I (inosine). Model selection can be automatic via modified base codes or explicit through model complexes. Newer kits and releases have expanded and refined model availability.

A practical starting command

Use a GPU-enabled call that includes multiple modified-base models. Dorado resolves the correct model pairing for your simplex model choice. The documentation describes how the --modified-bases flag accepts codes and selects the latest available model.

Where the calls live

• Modification calls are embedded in the BAM using MM and ML tags.
• MM tag encodes the base, strand, modification code, and positional offsets.
• ML tag records per-site probabilities as integers from 0 to 255, mapping evenly to 0.0–1.0.

These tags are defined in the SAM specification and remain tool-agnostic for downstream parsing.

From reads to transcripts

• The usual path moves from read-level calls to aggregated, transcript-level profiles.
• Convert BAM to FASTQ for alignment flexibility.
• Align reads to a transcript reference with minimap2.
• Run modkit pileup to summarize modification frequency into bedmethyl tables.

These tables support plotting, differential analysis, and downstream statistics.

Quality considerations

• Use motif-specific m6A models when your biology conforms to DRACH contexts.
• Keep kits and basecalling models consistent across cohorts when possible.
• Validate high-value sites with orthogonal assays or replicate evidence.

If you'd like to see a step-by-step walkthrough of basecalling commands, MM/ML tag interpretation, and transcript-level outputs, check out our guide to RNA methylation detection with direct RNA sequencing.

Step 4. Differential Methylation Analysis

Differences between conditions are often the real target. ONT's modkit dmr implements practical workflows for differential methylation loci (DML) and differential methylation regions (DMR). Use pair for two-condition comparisons that can evaluate single positions or user-defined regions. Use multi for region-level comparisons across more than two conditions.

Preprocessing matters

Compress and index bedmethyl files before scoring. Tabix indexing supports fast region queries and parallel scheduling. It also reduces IO overhead for large cohort studies.

Model and scoring

The DMR approach aims for simplicity and interpretability. It performs likelihood-ratio scoring under a model where potentially methylated sites within a region arise from the same distribution. This design enables exploratory analysis and rapid iteration across many regions or tracks.

Outputs to watch

• DML output reports locus-level differences, coverage, group proportions, and posterior metrics.
• DMR output aggregates over regions, producing difference scores, counts, and condition summaries.
• Balanced designs can include MAP-based p-values and paired effect sizes for replicate pairs.

Practical thresholds

• Set minimum coverage sensibly to avoid sparse, noisy sites.
• Evaluate effect size alongside probability to prioritize meaningful biology.
• Inspect top regions visually with Squigualiser where appropriate.

Putting It Together: A Reproducible DRS Playbook

A robust direct RNA sequencing analysis blends tool choice with disciplined practice. The following playbook has worked well for research teams and platform groups.

Data intake and storage

• Adopt a consistent folder structure for runs, barcodes, and batches.
• Store POD5 and metadata together with README files and checksums.
• Repack large POD5 files for better throughput before GPU basecalling.

Basecalling and modification models

• Pin Dorado versions per project and log exact arguments.
• Choose motif-specific models when they reflect your biology.
• Keep model updates synchronized across all study arms.

Visualization checkpoints

• Visualize a pilot region from each sample class.
• Confirm segmentation quality and alignment consistency.
• Record screenshots or HTML views for audit trails.

Aggregation and statistics

• Use modkit pileup to create consistent bedmethyl summaries.
• Filter by coverage and mapping quality before group comparisons.
• Run modkit dmr on positions or biologically relevant regions.

Reporting and review

• Summarize DML and DMR calls with effect sizes and coverage.
• Link sites to genes, isoforms, or functional motifs where possible.
• Cross-reference high-confidence hits with external knowledge bases.

Case-Oriented Scenarios You Can Reproduce

Plant stress biology

Profile seedling responses under drought and recovery conditions. Use the same chemistry and Dorado models across timepoints. Aggregate at the transcript level and evaluate m6A changes near stress-responsive isoforms. Visualize top regions that show motif-consistent shifts.

Cancer transcriptome exploration

Compare tumor and matched normal RNA using uniform basecalling. Focus on genes with regulatory roles and evaluate Ψ or m5C patterns where models support them. Weave in protein abundance data to test whether modifications associate with translation changes.

Method validation for a new kit

Select a small panel of housekeeping and modification-rich genes. Run both basecalling and visualization steps. Use likelihood-ratio scoring over canonical regions to quantify stability across replicates. Document findings in a concise validation report.

Common Pitfalls and How to Avoid Them

Inconsistent models across cohorts

Mixed model versions can generate artificial differences. Pin Dorado release and track full command lines in version-controlled notebooks.

Sparse coverage at long 3′ UTRs

APA and tail features complicate alignment and coverage. Set minimum coverage thresholds and consider region-level analyses for long transcripts.

Storage and IO bottlenecks

Massive POD5 stores can slow pipelines during peak usage. Repack files and use memory-mapped readers where possible. Plan for multi-terabyte storage with clear archival policies.

Ambiguous modification interpretation

Not all current shifts reflect the same chemistry. Confirm motifs, replicate consistency, and effect sizes before drawing conclusions. Add orthogonal validation for the most critical hits.

Tooling Notes and Ecosystem Bridges

SLOW5 and BLOW5 formats reduce overhead for random access workflows. They pair with slow5tools for conversion and validation. For POD5 conversion, the community blue-crab utility is recommended, and slow5tools documents the approach. These tools improve performance for signal-centric pipelines and collaborative environments.

SAM tag standards keep modified-base metadata interoperable. The MM and ML tags are formalized in the SAM specification. This standardization ensures that downstream parsers and libraries can remain tool-agnostic. It also promotes longevity of stored results.

Future Directions and Research Opportunities

Direct RNA sequencing will continue to expand its modification catalog. New Dorado models and chemistries will likely improve sensitivity and specificity. We also expect better handling of structured RNAs and dense modification contexts, including tRNAs. Recent work demonstrates pseudouridine mapping and joint detection of multiple modifications using DRS data. As models improve, study designs can explore richer questions with fewer compromises.

Integration with other omics should become routine. Pair DRS with ribosome profiling to study translation efficiency at modification-marked sites. Combine DRS with proteomics to link m6A or m5C status to protein abundance. Align findings with metabolomics to capture pathway-level responses. These combinations can move projects beyond association toward mechanism and intervention.

The standards layer will also mature. Continued adoption of the MM and ML tags will stabilize downstream packages. As more tools support these tags explicitly, it will be easier to move from basecalling to visualization and statistics without brittle glue code.

Conclusion

Direct RNA sequencing analysis provides a complete, modification-aware view of the transcriptome. The approach reads native RNA molecules, preserves poly(A) tails, and exposes chemical marks that guide regulation. A practical workflow starts with POD5 management, proceeds through visualization, and uses Dorado for modification calling. It then aggregates to transcript-level profiles and applies modkit dmr for differential analysis. These steps reveal meaningful differences between conditions while maintaining biological context.

Use this pillar page as an entry point for your content cluster. Publish the four deep-dive articles and link back here with clear anchor text. Keep URLs short, lowercase, and static. Maintain consistent formatting, model versions, and QC practices across the series. With that foundation, your site will serve both expert readers and search engines, while your teams gain a reliable guide for running direct RNA sequencing analysis end to end.

Direct RNA Sequencing Analysis: From Raw Signals to Differential Methylation

Direct RNA Sequencing – Signal File Handling and Visualization

Direct RNA Sequencing Analysis: Guide to RNA Modification and DMR Analysis