# Hebrew University researchers release BBERT to classify bacterial DNA in soil samples

Digest AI · Research · published 2026-10-29T15:58:51Z

Canonical: https://digestai.news/story/hebrew-university-researchers-release-bbert-to-classify-bacterial-dna

## Summary

Researchers at the Hebrew University of Jerusalem and partner institutions developed **BBERT**, an AI tool that identifies bacterial DNA in short genetic fragments as small as 100 bases. The tool addresses a long-standing challenge in environmental genomics: many sequenced soil samples contain unclassified genetic material due to incomplete reference databases, leading to what scientists call ‘microbial dark matter.’ BBERT uses a BERT-based model trained on DNA patterns to classify fragments as bacterial or non-bacterial, estimate protein-coding potential, and predict reading frames—without relying on reference sequences for each read. The model achieved about **90% accuracy** in distinguishing bacterial DNA, **88% accuracy** for identifying protein-coding fragments, and **96% accuracy** for determining reading frames in benchmark tests. It was tested on **1,969 publicly available soil DNA datasets**, improving assembly times from **66 hours to 25 hours** per dataset and reducing memory use in complex biological samples like legume root nodules. The tool is open-source and available for download, with support for GPU acceleration and compatibility with standard file formats like FASTA and FASTQ. While BBERT enhances efficiency and reduces computational load, researchers note limitations, such as potential misclassification of chloroplast DNA or difficulty distinguishing bacteria from archaea. The tool is intended to help scientists prioritize genetic fragments for deeper analysis, enabling faster and more targeted study of microbial communities in soil ecosystems.

## Key points

- BBERT classifies bacterial DNA in fragments as short as 100 bases with ~90% accuracy, per researchers
- Tool reduces soil DNA assembly time from 66 hours to 25 hours and cuts memory use in complex samples
- Open-source software supports GPU acceleration and integrates with standard genomic file formats

## Why it matters

BBERT accelerates microbial DNA analysis in soil, helping researchers uncover hidden microbial communities faster and more efficiently. This could advance studies of plant-microbe interactions, soil health, and environmental genomics, while reducing computational costs for large-scale datasets.

## Sources

1. [BBERT Uses AI to Recognize Bacterial DNA Hidden in Soil Samples](https://unite.ai/bbert-ai-bacterial-dna-soil-samples) (Unite.AI, 2026-10-29)

## Cite

Digest AI, "Hebrew University researchers release BBERT to classify bacterial DNA in soil samples", 29 October 2026, https://digestai.news/story/hebrew-university-researchers-release-bbert-to-classify-bacterial-dna

---

Written by Digest AI's editorial model from the linked sources; the sources are the record. Headlines, digests and key points are written by Digest AI and may be quoted with a link to the story page. Linked articles belong to their publishers. Terms: https://digestai.news/terms#reuse
JSON: https://digestai.news/story/hebrew-university-researchers-release-bbert-to-classify-bacterial-dna.json