New SRAminer tool mines NCBI’s Sequence Read Archive for hidden genes and microbes

The Sequence Read Archive at the US National Center for Biotechnology Information holds petabyte-scale records of raw sequencing data from studies around the world, but the archive was not designed for rapid, large-scale searching of its contents. Researchers report a new software tool, SRAminer, that searches and analyzes large sets of SRA datasets — up to 1,000,000 runs at a time — using user-defined keywords such as organisms, habitats, title words, sequencing strategy, project or platform. The tool then applies iterative BLAST searches, intensive computational analysis and iterative BLAST-assembly to reconstruct genes and genomes from reads that match a target of interest.

The developers describe the motivation behind the tool. Raw SRA data offers an enormous record of global living matter, yet it is not optimized for quickly finding and extracting matching sequences, such as genes, gene variants or hidden microbial taxa. SRAminer addresses that gap by searching and analyzing large sets of SRA datasets with user-defined keywords, then applying iterative BLAST searches and iterative BLAST-assembly to reconstruct genes and genomes from target-matched reads.

To demonstrate the tool, the developers ran four targeted searches across very different sample collections. In the largest, SRAminer screened 332,754 plant-parasitic nematode and rhizosphere samples and identified 92 SRA runs positive for Wolbachia, a widespread bacterial endosymbiont of invertebrates. A second search examined 945 hospital wastewater treatment plant samples and found 61 runs positive for pathogenic bacteria. A third, more focused query of 221 associated bacterial samples returned 3 runs positive for Salmonella enterica serovar Typhi, the typhoid fever agent. Finally, a sweep of 42,469 soil metagenomes uncovered 8 runs positive for Burkholderia pseudomallei, the soil-dwelling bacterium that causes melioidosis.

Each of these results illustrates the same principle: sequences of interest can be identified within large sets of archived datasets by searching the sequence reads themselves. By searching at the level of the sequence reads themselves, SRAminer can identify target-matched runs within the searched sample sets.

The technical pipeline behind the tool is iterative. After keyword-based selection narrows the field to relevant runs, BLAST searches compare target sequences against the reads, and matched reads are then fed into an assembly step that uses further BLAST-guided iterations to reconstruct longer stretches of the target gene or genome. This combination of keyword-based selection and sequence-level search and reconstruction is the approach the developers describe for finding genes and genomes of interest in the archive.

The scale of the demonstration searches is notable. The tool was designed to search and analyze large sets of SRA datasets of up to 1,000,000 runs, and the demonstration searches recovered rare targets, such as 3 positive runs among 221 associated bacterial samples, or 8 among 42,469 soil metagenomes. The larger yields, like 92 Wolbachia-positive runs from 332,754 plant-parasitic nematode and rhizosphere samples, show that the tool can identify target-matched runs within very large searched sample sets.

The developers anticipate that the tool will accelerate research into poorly studied genes and organisms in complex biological systems. The developers state that SRAminer is freely available, and they anticipate that the tool will accelerate research into poorly studied genes and organisms in complex biological systems. Potential applications span environmental and clinical microbiology, medicine and forensics, agriculture and food safety, and pathogen surveillance, according to the authors.

For pathogen surveillance in particular, the demonstration searches point to a practical pattern: hospital wastewater treatment plant samples yielded 61 pathogenic bacteria-positive runs, and soil metagenomes yielded 8 Burkholderia pseudomallei-positive runs, both from datasets that were already public. The authors anticipate that the tool could have widespread applications in environmental and clinical microbiology, medicine and forensics, agriculture and food safety, and pathogen surveillance.

SRAminer is described in BioData Mining as a methodology contribution, and the article is published open access. As with any mining approach, the identified runs represent sequence matches within the searched sample sets, and their biological interpretation will depend on the quality of the underlying data and the specificity of the target sequences used. The developers anticipate that the tool will streamline the exploration of massive sequence archives and enable efficient identification of genes and genomes of interest.

Subject of Research: A software tool for rapidly searching and extracting matching gene and genome sequences from NCBI's Sequence Read Archive.

Article Title: SRAminer: a tool for Sequence Read Archive data mining that rapidly finds sequence matches to hidden targets

Article References: Sharma, E., & Brown, A. M. V. (2026). SRAminer: a tool for Sequence Read Archive data mining that rapidly finds sequence matches to hidden targets. BioData Mining. https://doi.org/10.1186/s13040-026-00615-4

Image Credits: AI Generated

DOI: 10.1186/s13040-026-00615-4

Keywords: SRAminer, Sequence Read Archive, bioinformatics, BLAST, metagenomics, Wolbachia, Salmonella Typhi, Burkholderia pseudomallei, pathogen surveillance, data mining, genomics, environmental microbiology