Bioinformatics Advances Speed Up Genetic Research

Bioinformatics researchers at the universities of São Paulo and Campinas, in partnership with the UK's Wellcome Sanger Institute, published a study demonstrating how artificial intelligence tools can reduce by up to 60% the time required for large-scale genomic sequencing data analysis. The results, published in the journal Nature Methods, open concrete perspectives for accelerating research in rare diseases, precision oncology, and preventive medicine.

What the New Models Enable

The historical bottleneck of genomics is not data generation — modern sequencers produce gigabytes per hour — but its interpretation. Identifying clinically relevant genetic variants among billions of base pairs requires complex computational pipelines and specialists capable of contextualizing the results. The new AI models trained by the research group achieve filtering and prioritizing variants with precision comparable to that of experienced geneticists, in a fraction of the time.

In tests with a bank of 12,000 oncology patient genomes, the system correctly identified known pathogenic variants in 97.3% of cases and pointed out 214 variants of uncertain significance that conventional analyses had misclassified. This ability to reassess complex cases has direct implications for clinical diagnoses and for including patients in targeted therapy trials.

Democratizing Access and Next Steps

The team released the models as open-source software on GitHub, with documentation for integration into existing pipelines based on tools like GATK and ANNOVAR. University hospitals in Brazil and Portugal have already signaled interest in pilot tests. The researchers estimate that broader adoption of the tool could reduce the cost of diagnostic genomic analysis by up to 45%, making exams that currently cost thousands of reais more accessible to the public health system.