Method for Analyzing Molecules of Interest, Method for Automatically Selecting Electrical Signals Related to a Set of Known Barcode Molecules, Signal Processing Unit and Computer Program for Executing the Method

Simple SummaryContent extracted from patent full text and abstract with AI.

This patent describes a method and system for analyzing molecules, such as nucleic acids and proteins, using barcode adaptors that are attached to the molecules of interest and later identified by analyzing electrical signals generated during nanopore sequencing. The distinctive barcode sequence can be rapidly and accurately identified by segmenting and temporally aligning the electrical signal using dynamic time warping (DTW) and a machine learning classifier (Support Vector Machine, SVM), without the need for conventional basecalling or high-end computing resources. The method also covers the design of optimized barcode sets to maximize separation, improving multiplexing performance, speed, and accuracy, including in real-time sequencing runs.

Use CasesContent extracted from patent full text and abstract with AI.

• Multiplexed direct RNA or DNA nanopore sequencing, enabling simultaneous analysis of many samples in a single run.
• Single-cell transcriptomics and epitranscriptomics studies where accurate sample identification is critical.
• Diagnostics applications requiring rapid identification, quantification, or tracking of target molecules.
• Real-time selective sequencing (adaptive sampling), allowing dynamic enrichment of underrepresented samples or ejection of unwanted reads during sequencing.
• Protein or polymer analysis using nanopore or similar electrical detection methods, leveraging barcode adaptors.
• Optimization and quality control of sequencing experiments, including library preparation and barcode design workflows.

BenefitsContent extracted from patent full text and abstract with AI.

• Highly accurate demultiplexing of samples using electrical signal comparison, even with many (≥12) barcode types, enabling high-throughput workflows.
• Fast barcode identification (milliseconds per read) on standard CPUs, making the solution cost-effective and widely accessible, avoiding the need for GPUs.
• Minimally data-intensive – requires far fewer training reads than deep learning alternatives.
• Robust to signal variation and noise because of DTW-based alignment rather than fixed time windows or image transforms, giving high confidence even in challenging sequencing runs.
• Barcode set optimization provided by the method maximizes yield and accuracy, further improving multiplexing capacity.
• Method is independent of basecalling errors or basecaller updates, ensuring stable and reliable performance over time.
• Live (real-time) demultiplexing and adaptive control of sequencing runs, allowing dynamic resource allocation and enrichment strategies.
• Applicable to multiplexed sequencing of various molecule types (RNA, DNA, protein, polypeptide, or hybrids).
• Facilitates advanced experimental designs (e.g., single-cell, time-course, or viral kinetics studies) with improved accuracy, scalability and resource/cost efficiency.

Technical Classifications (CPCs)

Inventors & Applicants

Patent Abstract

A method for analyzing molecules of interest by use of an adaptor attached to the molecule of interest is described, wherein said adaptor comprises a barcode molecule, wherein the barcode molecule is one instance of a set of defined sequences of molecules. The method comprises the steps of: - partitioning the electrical signal derived by electrically measuring at least of the part of a molecule into segments, - temporarily aligning segments of the partitioned electrical signal related to an unknown barcode molecule derived by electrically measuring at least of the part of the molecule of interest comprising the barcode molecule and a set of respective signals for known barcode molecules stored in a data memory, and - predicting the instance of barcode molecule from the similarity between the temporarily aligned segments of the electrical signal and the respective signals for known barcode molecules, wherein the electrical signal is temporarily aligned by compressing or stretching the temporal axis and the instance of barcode molecule is predicted automatically in a data processing unit. The method is also directed to a process of barcode design based on the similarity of temporarily aligned segments of electrical signals related to the candidate barcode molecules and a set of designed target signals.