16 December 2025
Australian scientists have identified the key genetic drivers behind long COVID, revealing why some people continue to experience debilitating symptoms long after their initial infection.
The breakthrough, made using large scale biological datasets, could pave the way for targeted treatments and personalised diagnostics.
The team, led by University of South Australia scientists, integrated genetic and molecular data from more than 100 different international studies, identifying 32 causal genes that increase the likelihood of a person developing long COVID, including 13 new genes not previous associated with the disease.
Their findings have been reported in two new scientific papers published in PLOS Computational Biology and Critical Reviews in Clinical Laboratory Sciences.
An estimated 400 million people have been affected by long COVID since 2020, imposing a $1 trillion annual cost to the global economy.
Characterised by symptoms like prolonged fatigue, breathlessness, cardiovascular complications and cognitive impairment beyond four weeks, the condition has proved stubbornly difficult to diagnose and treat. Many people have experienced symptoms for weeks, months, and sometimes years after contracting the virus.
Lead author UniSA PhD candidate in Bioinformatics, Sindy Pinero, says large-scale datasets and advanced computational methods can more quickly identify the causes, risk factors, and potential treatment options for long COVID.
The methods combine advanced bioinformatics and artificial intelligence to interpret massive biological datasets known as “omics” data – encompassing genomics, proteomics, metabolomics, transcriptomics, and epigenomics.
“These findings mark a major step towards a more precise way of diagnosing and treating the condition,” Pinero says.
“Long COVID is incredibly complex. It affects multiple organs, shows highly variable symptoms, and has no single final diagnostic marker.
“However, by using computational models to integrate data from across the world, we can begin to uncover consistent molecular signatures of disease and identify biomarkers that point to new treatment targets.”
The review identifies dozens of genetic, epigenetic, and protein-level biomarkers linked to immune dysfunction, persistent inflammation, and mitochondrial and metabolic abnormalities.
Among the key discoveries is a genetic variant in the FOX P4 gene, associated with immune regulation and lung function, that appears to increase people’s susceptibility to long COVID.
Researchers also found 71 molecular switches that can turn genes on or off, persisting a year after infection, and more than 1500 altered gene expression profiles tied to immune and neurological disruption.
By integrating these findings using machine learning, the study demonstrates how different layers of biological data can be combined to predict which patients are at risk of long-term complications and how their symptoms may evolve.
“This computational framework not only improves our understanding of long COVID but could also accelerate the search for treatments for other post-viral symptoms such as chronic fatigue and fibromyalgia,” according to Assoc Prof Le.
Co-author, UniSA Associate Professor Thuc Le, says that computational science is essential to solving the long COVID puzzle.
“Traditional biomedical research can’t keep pace with the complexity of this condition,” Assoc Prof Le says.
“By applying artificial intelligence to global datasets, we can identify causal relationships that are invisible in small clinical trials – for example, how specific genes interact with immune pathways to drive persistent inflammation.”
The review also highlights the urgent need for larger, more diverse international datasets and longitudinal studies that follow patients for several years after infection.
“Many existing studies are small and inconsistent, which makes it hard to identify reliable biomarkers.
Global collaboration and data sharing are the key to producing results that can translate into clinical tools.
“This research is not only about long COVID. It represents a blueprint for how global science can use big data, AI and molecular biology to respond to future pandemics and complex chronic diseases.”
‘Integrative Multi-Omics Framework for Causal Gene Discovery in long COVID’ is published in PLOS Computational Biology DOI: 10.1371/journal.pcbi.1013725
‘Omics-based computational approaches for biomarker identification, prediction and treatment of long COVID’ is published in Critical Reviews in Clinical Laboratory Sciences (ILAB). DOI: 10.1080/10408363.2025.2583083
Media contact: Candy Gibson M: +61 434 605 142 E: candy.gibson@unisa.edu.au
Researcher contact: Sindy Pinero E: sindy.pinero@unisa.edu.au
