top of page

PtBio Releases Online Protocol for Creating Predictive AI Models Using DNA Language Models: 2026-04-15

  • Writer: PtBio Inc.
    PtBio Inc.
  • 2 hours ago
  • 2 min read

 PtBio Inc. (Headquarters: Higashi-Hiroshima City, Hiroshima Prefecture; CEO: Keisuke Okuhara; hereinafter “the Company”) is pleased to announce that its protocol paper, “Workflow for Fine-Tuning and Evaluating DNA Language Models for Specific Genomics Issues,” which outlines AI development protocols utilizing DNA language models to address specific genomics challenges, was published online on April 3, 2026, in the life sciences protocol journal “Bio-protocol” (ISSN: 2331-8325) (peer-reviewed). This paper was authored within the PtBio Joint Research Chair BioDX Laboratory at Hiroshima University, and the authors include Kazuki Nakamae, Team Leader of the BioDX Team in the Company’s Research and Development Department, and Hidemasa Bonou, Chief Scientific Officer (CSO).



A DNA language model is a type of artificial intelligence (AI) technology in which the AI learns from large amounts of DNA sequences composed of A, T, G, and C, much like how it learns the rules of language. In this protocol, we have presented, in a step-by-step and reproducible manner, the procedures for fine-tuning DNABERT-2—a representative example of such DNA language models—for specific research objectives and evaluating its performance. This protocol enables the construction of AI models capable of extracting meaningful features from DNA sequences without relying on the “intuition” or “personal expertise” of specialists.


In the paper, we used DNABERT-2 fine-tuning to predict “unintended effects on RNA (RNA off-target effects),” which are a major concern in genome editing. Furthermore, using a similar fine-tuning method, we predicted promoter sequences necessary for generating proteins responsible for biological functions from genetic information. We believe that such reproducible protocols can serve as a foundation for life science researchers to predict various genomic challenges and identify avenues for solving them.

Source: Nakamae, K. and Bono, H. (2026). Workflow for Fine-Tuning and Evaluating DNA Language Models for Specific Genomics Issues. Bio-protocol 16(9): e5676. DOI: 10.21769/BioProtoc.5676.
Source: Nakamae, K. and Bono, H. (2026). Workflow for Fine-Tuning and Evaluating DNA Language Models for Specific Genomics Issues. Bio-protocol 16(9): e5676. DOI: 10.21769/BioProtoc.5676.

 Guided by our vision of “Opening the Future with Biotechnology” and our mission to “Design biological functions to solve societal challenges,” we are developing platforms for genome editing and Bio-DX. This publication demonstrates our efforts to combine AI, genomic data, and computational infrastructure to create solutions that are practical for use in real-world research settings. Moving forward, we will continue to promote the societal implementation of AI-powered genomic analysis and genome editing in fields such as drug discovery and healthcare, food and agritech, and the environment and energy.


■Publication Information

Paper Title: Workflow for Fine-Tuning and Evaluating DNA Language Models for Specific Genomics Issues

Journal: Bio-protocol


■About PtBio Inc.

We are a startup established with the aim of bringing Hiroshima University’s cutting-edge biotechnology to practical application in society. As a platform provider that designs biological functions capable of solving global social challenges by leveraging our proprietary Bio-DX and genome editing technologies, we are promoting collaborative projects with various business partners.

Address: 10-23 Kagamiyama 3-chome, Higashi-Hiroshima City, Hiroshima Prefecture

Established: August 30, 2019

 Representative: Keisuke Okuhara, Representative Director and CEO

 Business Activities: Support for data-driven breeding utilizing Bio-DX and genome editing

 
 
bottom of page