Editorial: Artificial intelligence in genomic analysis
Frontiers in Bioinformatics · 06.11.2026 · 6
DOI: 10.3389/fbinf.2026.2036757 ↗Kurz gefasst
Fachartikel in der Zeitschrift Frontiers in Bioinformatics, online erschienen am 06.11.2026. Verfasst von João Carlos Setubal und 1 weiteren Autor:innen. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0).
Abstract (Originalsprache)
The next paper in the collec�on is by O. Abdelwahab and D. Torkamaneh (htps://www.fron�ersin.org/journals/bioinforma�cs/ar�cles/10.3389/�inf.2025.1694924/full). A�er poin�ng out the limita�ons of manual and heuris�c approaches to recognize gene�c variants in high-throughput sequencing data, a key step in the iden�fica�on of gene�c diversity, the authors propose a new framework, using the transformer deep neural network architecture, named VariantTransformer, in a binary classifica�on task: dis�nguishing between these variants from technical ar�facts. The authors report experiments comparing VariantTransformer with two alterna�ve pipelines using FASTQ files with sequence data. In the experiments carried out by the authors, the proposed framework presented the best predic�ve performance.(htps://www.fron�ersin.org/journals/bioinforma�cs/ar�cles/10.3389/�inf.2026.1718386/full) present a study using machine learning algorithms to induce models for phenotype predic�on. To overcome the complexity associated with the de novo assembly techniques for genome assembly, this study inves�gates the use of machine learning algorithms, in par�cular reinforcement learning algorithms. Reinforcement learning algorithms use the concepts of rewards and punishment to produce a model with high predic�ve performance. In the proposed approach, the authors train an agent using reinforcement learning to be able to iden�fy the correct order of reads from a sequenced genome. They propose new approaches for the reward step. These approaches are inves�gated through experiments with 23 datasets, and the results are assessed using distance from the expected genome, maximum reward and computa�onal cost.(htps://www.fron�ersin.org/journals/bioinforma�cs/ar�cles/10.3389/�inf.2025.1690229/full) apply machine learning techniques to RNA-seq data from Parkinson's Disease. The algorithms used were Logis�c Regression, Support Vector Machine, Random Forest, and XGBoost, complemente by network analysis with Cytoscape and the STRING database of protein-protein interac�ons. The authors found significant differences in gene expression between male and female pa�ents, and could point to some specific genes related to these differences. This study can be used as a model for RNA-seq-based inves�ga�ons of other complex disorders.In conclusion, we can make one firm statement: the use of AI in Genome Analysis is s�ll in its infancy. Coupled with the development of new instruments to probe and quan�fy molecular biology, AI can provide extraordinary new knowledge about the workings of life, followed by new approaches for biological func�on design. On the other hand, we wish to express the hope that the societal forces that are making possible this progress will also be able to prevent and act on the nega�ve side effects of AI, by adequate regula�on and containment.Author contribu�ons JCS: Wri�ng -review and edi�ng, Wri�ng -original dra�. ACPLFC: Wri�ng -review and edi�ng.
Zitieren
APA
Setubal, J. C., & Carvalho, A. C. P. D. L. F. D. (2026). Editorial: Artificial intelligence in genomic analysis. Frontiers in Bioinformatics, 6. https://doi.org/10.3389/fbinf.2026.2036757
Vancouver
Setubal JC, Carvalho ACPDLFD. Editorial: Artificial intelligence in genomic analysis. Frontiers in Bioinformatics. 2026;6. doi:10.3389/fbinf.2026.2036757
BibTeX
@article{Setubal2026,
author = {Setubal, João Carlos and Carvalho, André Carlos Ponce de Leon Ferreira de},
title = {Editorial: Artificial intelligence in genomic analysis},
journal = {Frontiers in Bioinformatics},
year = {2026},
volume = {6},
doi = {10.3389/fbinf.2026.2036757}
}
EUVIMED