← Latest papers
💻 bioinformatics

HARICOT: De Novo Assembly of Raw Nanopore Signals using Fuzzy de Bruijn Graphs

HARICOT is a novel tool that improves the de novo assembly of raw nanopore signals by employing fuzzy de Bruijn graphs and Hough-transform-based overlap detection to overcome the noise sensitivity and scalability limitations of existing methods, achieving contiguity and quality comparable to leading nucleotide-based assemblers without requiring basecalling.

Original authors: McConnell, U., Ngo, Q., Firtina, C.

Published 2026-11-18
📖 4 min read☕ Coffee break read

Original authors: McConnell, U., Ngo, Q., Firtina, C.

Original paper licensed under CC BY 4.0 (https://creativecommons.org/licenses/by/4.0/). ⚕️ This is an AI-generated explanation of a preprint that has not been peer-reviewed. It is not medical advice. Do not make health decisions based on this content. Read full disclaimer

Imagine trying to read a book where the letters are not printed on the page, but are instead written in a shifting, wavy line of ink that changes thickness and color as you move your finger along it. This is the reality of modern DNA sequencing using a technology called nanopore sequencing. As a strand of DNA passes through a tiny hole, it disrupts an electrical current, creating a raw signal—a jagged, noisy waveform that represents the genetic code. Traditionally, scientists have had to translate this messy wave into the familiar letters of the genetic alphabet (A, C, G, T) before they could do anything useful with it. This translation step, known as basecalling, is slow and requires massive computing power, often relying on complex artificial intelligence models. But what if you could skip the translation entirely and assemble the story of the genome directly from the raw, noisy waves?

For years, researchers have attempted to piece together these raw signals without converting them first, hoping to save time and computing resources. One previous tool, called Rawsamble, managed to find overlaps between these signals, but the resulting genetic maps were often broken into tiny, useless fragments, much like a puzzle where most of the pieces are missing or misshapen. The problem was that the tools used to put these pieces together were designed for clean, translated text, not for the chaotic, error-prone nature of raw electrical signals. The noise in the signals made it nearly impossible for standard methods to find the correct connections, leading to a fragmented and incomplete picture of the genome.

A team of researchers has now introduced a new method called HARICOT that changes how we approach this problem. Instead of trying to force the noisy signals into a format they were never meant for, HARICOT embraces the noise. It uses a mathematical technique borrowed from computer vision, originally designed to find straight lines in a scatter of random dots, to identify where two different DNA signals overlap. By treating the raw electrical waves as a pattern of points rather than a sequence of letters, the system can find the hidden lines of connection that other tools miss. Once these overlaps are found, the system builds a "fuzzy" map of the genome. This map allows for slight imperfections and variations, grouping together signal segments that are similar but not identical, effectively smoothing out the errors that would otherwise break the assembly.

The results of this new approach are striking. When tested on various genomes, from simple bacteria to complex plants and fruit flies, HARICOT produced genetic maps that were significantly longer and more continuous than those created by the previous best tool. In many cases, the new method assembled entire bacterial chromosomes into single, unbroken pieces, a feat that previously required translating the signals into letters first. While the new tool still makes a few more mistakes than the most advanced methods that do use translation, it achieves a level of completeness that was thought to be impossible without that translation step. Perhaps most importantly, it does all of this much faster than the old method, running nearly four times faster on average and up to eleven times faster on larger genomes, all while avoiding the heavy computational cost of translating the signals into letters.

This work suggests that we do not always need to convert raw biological data into a standard format to understand it. By developing tools that can work directly with the messy, real-world signals, scientists can potentially analyze genomes in real-time as they are being sequenced, without waiting for slow processing steps. The researchers have made their code and models available to the public, inviting others to build on this foundation. While challenges remain, particularly with very complex genomes and certain types of signal errors, HARICOT demonstrates that it is possible to navigate the noise of raw data and find the clear, continuous story of life hidden within.

Drowning in papers in your field?

Get daily digests of the most novel papers matching your research keywords — with technical summaries, in your language.

Try Digest →