JERUSALEM, Oct. 29 (Xinhua) -- Israeli researchers have developed an artificial intelligence (AI) tool that can identify bacterial DNA in tiny genetic fragments found in soil, the Hebrew University of Jerusalem said Thursday.
Researchers extract DNA from soil samples to study the microorganisms living there, but identifying the source of many genetic fragments remains difficult.
The tool, called BBERT, is an AI model designed to process highly fragmented DNA/RNA sequences as short as 100 DNA letters and learn patterns in bacterial DNA, the university said in a statement.
In their study, published in Nucleic Acids Research, the researchers showed that the tool can identify bacterial DNA, including genetic material that can't be matched to existing reference databases.
The tool helps distinguish bacterial DNA from that of other organisms and identify fragments containing protein-making instructions, they said.
In tests, BBERT distinguished bacterial from non-bacterial DNA with around 90 percent accuracy, identified protein-making fragments with 88 percent accuracy and their correct reading frames with 96 percent accuracy.
Tested on 1,969 publicly available soil DNA datasets, the tool identified many fragments that existing methods had left unclassified.
The researchers said that the tool could help uncover previously overlooked genetic information and better understand the diversity of life in soil. ■
