A newly published collection of studies demonstrates how artificial intelligence techniques are pervading genome analysis and the study of omics datasets, while researchers emphasize that the integration of artificial intelligence into this field remains in its infancy. Editors João Carlos Setubal and André Carlos Ponce de Leon Ferreira de Carvalho note that this technological trend aligns with broader initiatives across science, such as the 2026 Strategic Plan of the American Department of Energy Joint Genome Institute, titled “AI-Driven Genomics to Accelerate Biological Discovery.” Within the curated Research Topic in Frontiers in Bioinformatics, six featured investigations illustrate how computational models address complex challenges in sequencing, variant identification, phenotype prediction, and disease expression.
Highlighting the scope of these technologies in human health, Wellington Rodrigues and colleagues conducted a systematic scoping review analyzing 1,040 articles from the scientific literature. Their synthesis revealed that artificial intelligence has become a driving technique within genome analysis, providing an established taxonomy of computational methods alongside timelines marking their historical introduction and practical application across biological research.
Despite the widespread adoption identified in the review, Rodrigues and coauthors determined that translating artificial intelligence into clinical practice and public health contexts remains contingent upon several critical methodological benchmarks. The authors established that successful clinical translation will depend directly on stronger external validation, clearer reporting standards, improved model calibration, biologically grounded interpretability, more inclusive datasets, and explicit governance frameworks to oversee analytical implementations.
Addressing technical challenges in sequencing accuracy, Jinnan Hu and colleagues presented the DNAscope Hybrid variant calling pipeline, an analytical method designed to achieve high precision by integrating short-read and long-read sequencing data. The authors structured the hybrid strategy to leverage the distinct advantages of each sequencing modality while mitigating inherent individual disadvantages, seeking enhanced accuracy alongside optimized computational and sequencing costs.
The pipeline developed by Hu and collaborators targets multiple classes of genetic variation, specifically Single Nucleotide Variants, short insertions and deletions of less than 50 base pairs, and longer structural variants. Machine learning models perform a relatively minor role within this framework compared to other entries in the collection, as the majority of the pipeline’s analytical results stem from the efficient implementation of established mapping algorithms.
To resolve limitations inherent to manual and heuristic approaches for detecting sequence variation in high-throughput sequencing data, Abdelwahab and Torkamaneh introduced a deep neural network framework named VariantTransformer. Utilizing a transformer architecture, the framework executes a binary classification task designed to separate true genetic variants from technical sequencing artifacts, representing an essential stage in characterizing biological diversity. When evaluated experimentally against two alternative pipelines analyzing FASTQ sequence files, VariantTransformer delivered superior predictive performance.
The investigators trained classifiers on Single Nucleotide Polymorphism sequences to predict resistance to Viral Nervous Necrosis across an experimental dataset comprising 990 young European sea bass specimens from previously published research.
As part of their modeling approach, Faldani and coauthors implemented convolutional neural networks by converting Single Nucleotide Polymorphism sequences into graphical images.
Targeting the computational complexity of de novo genome assembly, Padovani and colleagues evaluated machine learning methodologies, focusing specifically on reinforcement learning algorithms that optimize performance through mechanisms of rewards and punishments. The researchers trained an agent using reinforcement learning principles to determine the correct ordering of sequence reads from a sequenced genome while proposing new approaches for the reward evaluation step, assessing the system across 23 distinct datasets based on distance from the expected genome, maximum reward scores, and computational cost.
In an application to complex human neurological disorders, Perperidis and coauthors applied machine learning algorithms—including Logistic Regression, Support Vector Machines, Random Forests, and XGBoost—to Parkinson’s Disease RNA-seq datasets, combining these models with network analysis via Cytoscape and the STRING protein-protein interaction database. Their analysis identified significant differences in gene expression between male and female patients, highlighting specific genes associated with these disparities to serve as a template for RNA-seq studies in other disorders; in concluding the topic, the editors emphasized that while AI tools coupled with molecular instrumentation offer new opportunities for biological function design, appropriate containment and regulation will be essential to mitigate potential societal risks.
The Research Topic, titled “Artificial intelligence in genomic analysis,” was curated within the Genomic Analysis section of Frontiers in Bioinformatics by João Carlos Setubal of the Instituto de Química at the Universidade de São Paulo and André Carlos Ponce de Leon Ferreira de Carvalho of the Instituto de Ciências Matemáticas e de Computação at the Universidade de São Paulo in São Carlos, with editorial review conducted by Shandar Ahmad of Jawaharlal Nehru University in India. Following submission on 21 October 2026 and acceptance on 26 October 2026, the editorial was published on 06 November 2026 under an open-access Creative Commons Attribution License.
Subject of Research: Applications of artificial intelligence in genomic analysis
Article Title: Editorial: Artificial intelligence in genomic analysis
Article References: Setubal, J. C., & de Carvalho, A. C. P. D. L. F. (2026). Editorial: Artificial intelligence in genomic analysis. Frontiers in Bioinformatics, 6, Article 2036757. https://doi.org/10.3389/fbinf.2026.2036757
Image Credits: AI Generated
DOI: 10.3389/fbinf.2026.2036757
Keywords: genomic analysis, artificial intelligence, machine learning, bioinformatics, variant calling, deep learning, reinforcement learning, transcriptomics, Editorial, Artificial, intelligence, genomic
News Source: Blake Davidson. (November 7, 2026). Artificial Intelligence Applications Advance Genomic Analysis Across Diverse Biological Studies. Scienmag.
