Cancer mutations are alterations in the DNA of tumour cells that are absent from healthy tissue. Finding these changes accurately is essential for understanding how cancers develop, tracking disease progression and selecting treatments tailored to individual patients. Yet the task is technically demanding. Tumour samples often contain a mixture of cancerous and normal cells, while mutations can occur in repetitive or structurally complex sections of the genome that are difficult to reconstruct from short fragments of DNA.

Most existing somatic-variant callers—the software tools used to distinguish tumour mutations from inherited genetic differences—were created primarily for short-read sequencing. Short-read platforms produce large numbers of highly accurate fragments, but each fragment covers only a small portion of the genome. Long-read sequencing, by contrast, generates much longer DNA molecules that can span repetitive sequences, structural rearrangements and other difficult regions. This broader view can reveal genomic changes that would otherwise remain hidden, although it also creates new computational challenges.

ClairS tackles these challenges with a neural-network architecture trained to interpret the complex signals produced by long-read tumour-normal sequencing. The system compares DNA data from a tumour with a matched normal sample and searches for small somatic variants, including single-nucleotide changes and short insertions or deletions. Rather than relying only on fixed rules, the model learns patterns associated with genuine tumour mutations, sequencing errors and differences caused by the proportion of cancer cells present in a sample.

One of the most innovative aspects of ClairS is the way its developers generated training data. High-quality tumour-normal datasets are scarce, expensive to produce and difficult to obtain in sufficient quantities. To overcome this limitation, the researchers mixed sequencing data from normal human samples to create synthetic tumour-normal pairs. The process allowed them to simulate a wide range of biological and technical conditions, including different tumour purities, sequencing depths and mutation burdens.

This synthetic-data strategy gives the model access to an effectively unlimited supply of realistic training examples. Tumour purity is particularly important because a mutation may appear in only a small fraction of the DNA molecules analysed. If the cancer cells represent a minor component of a biopsy, the signal from a true mutation can be overwhelmed by normal DNA. By exposing ClairS to simulated samples with varying levels of tumour purity, the researchers trained it to recognise weak but meaningful mutation signals under conditions that resemble real clinical specimens.

The team evaluated ClairS using datasets from several cancer types, including breast cancer, lung cancer, melanoma and pancreatic cancer cell lines. Across different sequencing conditions, the algorithm showed high accuracy in detecting small somatic mutations. Its performance was particularly important in regions where long reads provide an advantage, because the extended DNA fragments can preserve the genomic context needed to distinguish a true mutation from a technical artefact.

Unlike many experimental algorithms that remain confined to academic demonstrations, ClairS has already been incorporated into the official somatic-variant-calling workflow of Oxford Nanopore Technologies. The integration places the method inside a practical commercial analysis pipeline and could speed its adoption by researchers and clinical genomics laboratories. Although further validation will be needed before any tool is used routinely for patient diagnosis or treatment decisions, the development represents a significant step toward making long-read cancer analysis more accessible.

“Long-read sequencing is transforming how we study cancer genomes, especially in regions that were previously difficult to analyse,” said Professor Ruibang Luo, the study’s senior researcher and an Associate Professor at HKU’s School of Computing and Data Science. “ClairS makes it possible to train powerful AI models even when real cancer training data is limited, supporting more reliable cancer mutation discovery from long-read sequencing data.”

The work also illustrates a broader shift in biomedical AI. Many medical algorithms are limited not by a lack of computational power, but by the shortage of accurately labelled clinical data. ClairS demonstrates how carefully designed simulations can provide a practical bridge between limited real-world samples and the enormous diversity of conditions encountered in biology. By combining long-read sequencing with deep learning and scalable synthetic-data generation, the method could help researchers build more complete cancer genomes, uncover mutations missed by traditional approaches and advance the development of precision oncology. The study, published in Nature Methods, is open source, allowing the wider genomics community to inspect, reproduce and further develop the technology.

Subject of Research: Computational simulation/modeling

Article Title: ClairS: a deep-learning method for long-read tumor–normal pair somatic small variant calling

News Publication Date: 1 July 2026

Web References: https://github.com/HKU-BAL/ClairS

References: Nature Methods. DOI: 10.1038/s41592-026-03152-4

Image Credits: The University of Hong Kong

Keywords: ClairS, cancer mutations, long-read sequencing, deep learning, artificial intelligence, somatic variant calling, tumour genomics, precision medicine, bioinformatics, Oxford Nanopore Technologies

The central obstacle is access: conventional antibodies are generally unable to reach intracellular targets. To overcome this, the team exploited the way cells naturally process proteins. When mutated KRAS(G12D is broken down, it can generate short protein fragments—neoantigens—that act as molecular “clues” for the immune system.

In many cases, such neoantigen fragments are loaded onto the cell surface for immune surveillance. The researchers focused on designing an antibody that can “read” one of these KRAS(G12D)-derived neoantigen fragments. Using a computational-to-experimental workflow, they selected candidates that would bind only to cancer cells presenting the relevant mutation-derived trace.

Their antibody is described as TCR-like, borrowing design logic from T-cell receptors, which recognize peptide fragments displayed on the major histocompatibility complex. In effect, the antibody provides an immunotherapy molecule with an analog of the T cell’s sensing mechanism, enabling it to distinguish cancer-associated intracellular mutations from normal cellular proteins.

Experimental tests showed that the antibody selectively binds KRAS(G12D)-bearing cancer cells while exhibiting minimal reactivity to non-mutant targets. Functional assays further indicated that the antibody can eliminate mutation-positive cancer cells in immunotherapy settings, supporting the concept that intracellular driver mutations can be made therapeutically visible.

Importantly, the study presents its platform as more than a single-mutation achievement. Because neoantigen generation is a general feature of mutated proteins, the same design strategy could be adapted to other cancer mutations that generate distinct intracellular fragments.

The research, led by Professor Byung-Ha Oh of KAIST’s Department of Biological Sciences with collaboration from Therazyne, was conducted by KAIST-affiliated investigators including SangPhil Ahn at Therazyne. The paper was published online in Molecular Therapy, reflecting the journal’s focus on gene and cell therapy innovations.

Overall, the study positions computational protein design paired with targeted screening as a route to next-generation precision antibody therapies. By aiming specificity at mutation-derived neoantigen signatures, the approach seeks to improve therapeutic discrimination and reduce collateral effects on healthy cells.

Subject of Research: TCR-like antibody targeting the KRAS(G12D) neoantigen (intracellular cancer mutation)
Article Title: Discovery of TCR-like antibodies to the KRAS G12D neoantigen via in silico-in vitro workflow
News Publication Date: 24-Jul-2026
Web References: http://dx.doi.org/10.1016/j.ymthe.2026.05.032
References: 10.1016/j.ymthe.2026.05.032
Image Credits: Credit: KAIST

Keywords: KRAS(G12D), neoantigen, TCR-like antibody, computational protein design, in silico-in vitro workflow, precision immunotherapy, intracellular targets, Molecular Therapy