Researchers at Baylor College of Medicine and the Jan and Dan Duncan Neurological Research Institute at Texas Children’s Hospital have developed a new method to improve the accuracy of genetic testing. Their work, published in Nature Communications, introduces local ancestry inference (LAI) as a way to analyze genetic variants in the Genome Aggregation Database (gnomAD), which is widely used for studying human genetics.
The LAI method divides the genome into segments based on specific ancestries, allowing researchers to better understand differences in genetic variation. Dr. Elizabeth Atkinson, assistant professor in the Department of Molecular and Human Genetics at Baylor and principal investigator at the Duncan NRI, explained, “This research updates our genomic resources to reflect the full spectrum of genetic variation. By refining allele frequency estimates for admixed populations, we can improve the accuracy of genetic diagnoses and reduce the risk of misclassification — ultimately benefitting patients across all backgrounds."
The study titled “Improved Allele Frequencies in gnomAD through Local Ancestry Inference” marks progress in personalized medicine by providing more precise data for interpreting genetic tests. Atkinson served as senior author, with Pragati Kore and Michael Wilson as co-first authors.
Genetic testing relies on understanding how common certain variants are within populations; if a variant is widespread among healthy individuals, it is likely benign. However, current approaches often use averages across large groups without accounting for people with mixed ancestry backgrounds. This can hide important differences that may affect diagnosis.
By applying LAI, Atkinson’s team calculated how frequent each variant was within different ancestry segments such as African, European or Indigenous American origins. They found that many variants considered rare globally were actually common within specific ancestral groups.
“These differences are not just academic,” said Atkinson. “They have clinical consequences.” The research showed that over 80% of genetic sites in African/African American and Latino/Admixed American groups had higher frequencies in at least one ancestry-specific segment than previously reported. In some cases, this meant reclassifying variants from potentially harmful to benign based on thresholds set by the American College of Medical Genetics and Genomics.
The updated ancestry-specific data has been made publicly available through gnomAD so that researchers and clinicians can access more detailed information when interpreting genetic results.
“Ancestry is a complex, and putting a single label on patients is not the most accurate way to diagnose them,” Atkinson said. “With this research, we are moving toward a more nuanced consideration of ancestry.”
