From sequencing to intelligence: how AI is transforming metagenomics

Microbial communities are critical in advancing human health. Metagenomics is a technique that analyzes these communities and allows for investigating their composition and functions. Metagenomic shotgun sequencing enables to capture all of the genetic material in environmental samples, such as water, soil, or the human gut. Despite this advantage, one of the main challenges of this technique is the assembling and interpreting of its data, as it produces many short, fragmented reads. Though long-read technologies may change this in the future, artificial intelligence (AI), machine learning (ML) and data science (DS) offer a powerful solution now, enabling scientists to efficiently process and analyze these large and complex datasets. This review explores the latest advancements in AI and ML applications across the metagenomic pipeline. First, it examines the impact of deep learning (DL) on next-generation sequencing, particularly for long-read technologies. Then, it discusses how ML is automating and improving quality control processes, as well as the use of AI applications in metagenome-assembled genome (MAG) assembly, with a focus on contig binning. Finally, this article looks at how AI and ML can improve predictive modeling for phenotype prediction.