DeepRM – Nanopore RNA Modification Detection Implementation Service

DeepRM is a professional implementation service for researchers and bioinformatics teams working with Oxford Nanopore direct RNA sequencing data. Rather than building and maintaining your own pipeline, we configure and deploy a deep learning-based RNA modification detection workflow into your AWS account via AWS HealthOmics.

Your team runs analyses directly within your own AWS environment, with all infrastructure costs billed to your account under standard AWS pricing.

Upon purchase, our team will:

• Share the pre-configured AWS HealthOmics workflow to your AWS account
• Guide you through input data requirements and preparation
• Walk you through job submission and parameter configuration
• Provide technical support for troubleshooting and result interpretation

The workflow we configure and deploy consists of two sequential stages, executed within AWS HealthOmics:

Stage 1 – Preprocessing Prepares input data for model inference by performing signal preprocessing and data normalization using your provided POD5 and BAM files. Required inputs:

• POD5 files: raw electrical signal data from Nanopore direct RNA sequencing
• BAM files: coordinate-sorted output from the Dorado basecaller, with BAM index (.bai) files

Stage 2 – Inference Runs GPU-accelerated deep learning inference for RNA modification detection. GPU compute instances are automatically provisioned by AWS HealthOmics.

Before initiating a run, you provide the following Amazon S3 paths:

• POD5 raw signal files
• Coordinate-sorted BAM file (Dorado basecaller output)
• BAM index (.bai) file
• Output destination S3 path

AWS HealthOmics automatically provisions and manages all compute infrastructure for each run. No instance selection or environment configuration is required.

Stage 1 – Preprocessing

• CPU: 192 vCPUs
• Memory: 384 GiB

Stage 2 – Inference

• CPU: 48 vCPUs
• Memory: 384 GiB
• GPU: 4 × NVIDIA L40S GPUs
• Total GPU Memory: 96 GiB

Results are saved to your designated Amazon S3 path:

• Per-read RNA modification probability scores
• Site-level RNA modification summary file with stoichiometry estimates

Outputs are compatible with standard bioinformatics pipelines, visualization tools, and downstream statistical analysis workflows.

In internal testing, end-to-end analysis of 8 POD5 sample files (113 GB total) completed in approximately 5 hours. Actual runtime varies based on dataset size, alignment complexity, and input characteristics.

DeepRM was used to construct a comprehensive human m6A atlas at single-molecule resolution, identifying a large number of previously uncharacterized non-canonical m6A sites and differentially modified transcripts across the human transcriptome.

Reference: Comprehensive discovery of m6A sites in the human transcriptome at single-molecule resolution  (Nature Communications, 2025)

This professional services listing has no software license fee.

⚠️ Important: Running the DeepRM workflow incurs AWS infrastructure charges billed directly to your AWS account, separate from this Marketplace listing. These include:

• AWS HealthOmics service fees
• EC2 compute costs (CPU and GPU instances)
• Amazon S3 storage and data transfer costs Customers are responsible for all AWS service charges incurred during workflow execution.

Pre-purchase: Contact us at  support@genome4me.com  to discuss dataset requirements, estimated run costs, or to request a sample analysis.

Post-purchase: We respond to all support requests within 2 business days. Support covers workflow onboarding, job submission, troubleshooting, and result interpretation guidance.

• Epitranscriptomics researchers
• RNA modification site discovery projects
• Single-molecule RNA modification profiling studies
• Teams quantifying RNA modification stoichiometry from Nanopore direct RNA sequencing data