Related Experiment Video
Updated: Jun 26, 2026
13:24
Integration of Wet and Dry Bench Processes Optimizes Targeted Next-generation Sequencing of Low-quality and Low-quantity Tumor Biopsies
Published on: April 11, 2016
11.8K
EpiDiP/NanoDiP: a versatile unsupervised machine learning edge computing platform for epigenomic tumour diagnostics
Jürgen Hench1, Claus Hultschig2, Jon Brugger2
1Institut für Medizinische Genetik und Pathologie, Universitätsspital Basel, Schönbeinstr. 40, 4031, Basel, Switzerland. juergen.hench@usb.ch.
Acta Neuropathologica Communications
|April 4, 2024
Summary
A new open-source suite, EpiDiP/NanoDiP, offers precise DNA methylation and copy number profiling for diagnostic tumor typing across various sequencing methods. It operates offline on edge devices, ensuring data privacy and enabling rapid intraoperative classification.
Area of Science:
- Epigenetics and Molecular Diagnostics
- Computational Biology and Bioinformatics
- Oncology and Pathology
Background:
- DNA methylation profiling is crucial for precise tumor typing, with microarrays and nanopore sequencing being common methods.
- Existing tools lack comprehensive support for diverse methylation profiling data (microarray, nanopore, parallel sequencing) and offline operation.
- There's a need for integrated software that can handle various data types, control sequencers, and run in resource-limited diagnostic labs.
Purpose of the Study:
- To introduce EpiDiP/NanoDiP, an open-source suite for DNA methylation and copy number profiling.
- To provide a versatile tool for classifying and interpreting data from microarray, nanopore, and parallel sequencing.
- To enable offline, privacy-preserving analysis on edge computing devices for rapid diagnostic tumor classification.
Main Methods:
- Development and benchmarking of the EpiDiP/NanoDiP open-source DNA methylation and copy number profiling suite.
- Utilized in-house routine diagnostic data from 2019-2021 for benchmarking against established supervised machine learning approaches.
- Designed for local operation on cost- and energy-saving edge computing devices (System-on-Chip, gpGPU-augmented) in offline mode.
Main Results:
- EpiDiP/NanoDiP successfully classifies and interprets DNA methylation and copy number data from diverse sequencing platforms.
- The software operates effectively in offline mode on edge devices, ensuring data privacy and enabling rapid analysis.
- Benchmarking confirmed its performance against established methods, with potential for intraoperative molecular tumor classification.
Conclusions:
- EpiDiP/NanoDiP addresses the limitations of existing tools by offering a unified, versatile platform for DNA methylation profiling.
- The suite facilitates rapid, privacy-preserving tumor classification, supporting diagnostic routine and epigenetics research.
- It serves as a valuable resource for pathologists and researchers, enhancing diagnostic precision and efficiency.
