← Latest papers
💻 bioinformatics

LDEN: Alternating projections for whole genome linkage disequilibrium estimation from next-generation sequencing data

This paper introduces LDEN, a novel, population reconstruction-independent method that uses alternating projections to efficiently and accurately estimate whole-genome linkage disequilibrium from next-generation sequencing data by formulating the problem as a matrix completion task constrained by positive semidefiniteness.

Original authors: Abdullah, S. M. U., Sun, Y.

Published 2026-11-17
📖 5 min read🧠 Deep dive

Original authors: Abdullah, S. M. U., Sun, Y.

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 ⚕️ This is an AI-generated explanation of a preprint that has not been peer-reviewed.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

Viruses are masters of disguise and rapid change. Unlike humans, who pass down a single, stable set of instructions to their offspring, viral populations often exist as a chaotic swarm of slightly different versions, known as variants. Within this swarm, some combinations of genetic changes work together to make the virus stronger, while others might cancel each other out or even weaken it. To understand how a virus evolves and which mutations are truly dangerous, scientists need to see how these genetic changes are linked. If two mutations always appear together, they are said to be in "linkage disequilibrium," a term that simply means they are hitching a ride on the same genetic background. Knowing this connection is crucial for distinguishing a mutation that is genuinely harmful from one that just happens to be traveling with a harmful partner.

The problem is that the tools scientists use to read viral DNA, called next-generation sequencing, have a built-in blind spot. These machines chop the viral genome into tiny fragments to read them, but the fragments are often much shorter than the full viral blueprint. Imagine trying to understand the plot of a long novel by reading only short, isolated paragraphs; you can see the words in a single paragraph, but you cannot tell if a character mentioned in the first sentence is the same one appearing in a paragraph ten pages away. Because the reading fragments are so short, scientists can easily see which mutations appear together on the same tiny piece of DNA, but they cannot see which mutations are linked across the rest of the genome. This missing information has made it difficult to map the true evolutionary relationships within viral populations, especially for large viruses like HIV or norovirus.

To solve this puzzle, researchers Syed Muhammad Umer Abdullah and Yanni Sun developed a new method called LDEN, which stands for whole genome linkage disequilibrium estimation from next-generation sequencing data. Instead of trying to rebuild the entire viral genome piece by piece—a process that is slow, memory-hungry, and prone to errors when data is sparse—the team approached the problem as a mathematical completion task. They realized that while the sequencing data only shows a small, visible slice of the genetic connections, the entire map of connections must follow a specific mathematical rule: it must be a "positive semidefinite" matrix. In plain language, this means the map of genetic links must be internally consistent and cannot contain impossible contradictions.

The researchers treated the known genetic links found in the short DNA fragments as fixed points on a map and used the mathematical rule of consistency to fill in the missing gaps. They built an algorithm that repeatedly adjusts the unknown parts of the map, forcing them to fit the known data while ensuring the whole structure remains mathematically valid. This approach allowed them to infer the invisible connections between distant mutations without ever having to reconstruct the full viral sequences. By doing so, they bypassed the need to identify every single variant in the population, focusing instead directly on the relationships between the mutations themselves.

The results of this new approach were striking. When tested on simulated viral populations, the method proved to be incredibly fast, completing the analysis in about ten seconds, which is roughly one hundred times faster than previous methods that relied on reconstructing the full population. Despite this speed, the accuracy was higher. The new method successfully recovered valid genetic maps that older methods often failed to produce, particularly in regions where the data was thin or noisy. The researchers also found that using paired-end sequencing, where both ends of a DNA fragment are read, provided even more information, further sharpening the accuracy of the genetic map without slowing down the process.

To prove the method worked on real-world biology, the team applied it to data from a patient infected with norovirus, a virus that causes severe stomach illness. The patient was undergoing treatment with a drug called favipiravir, and the virus had developed multiple mutations during the infection. By using LDEN to map the genetic links between these mutations, the researchers could calculate which mutations were being strongly selected for by the virus's environment. The analysis revealed that the mutations located in the parts of the virus that the immune system targets were indeed the ones under the strongest pressure to change. This finding aligned perfectly with biological expectations, confirming that the method could uncover meaningful evolutionary insights from complex patient data.

The study demonstrates that it is possible to see the full picture of viral evolution without needing to reconstruct every single detail of the genome. By relying on the mathematical consistency of the data rather than brute-force reconstruction, LDEN offers a faster and more reliable way to understand how viruses adapt. This capability is vital for tracking how viruses like HIV, influenza, and SARS-CoV-2 evolve, potentially helping scientists identify dangerous variants sooner and understand how they might respond to treatments. The work suggests that with the right mathematical framework, the limitations of current sequencing technology can be overcome, turning fragmented data into a clear view of the viral world.

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 →