Beyond sequence similarity: toward function-based screening of nucleic acid synthesis.

Area of Science:

• Biotechnology
• Biosecurity
• Bioinformatics

Background:

• Synthetic nucleic acids are crucial for biotechnology but pose dual-use risks.
• Current screening relies on sequence similarity, which may fail against AI-generated sequences.
• AI can decouple biological function from known sequences, evading detection.

Purpose of the Study:

• To highlight the need for function-based screening of synthetic nucleic acids.
• To assess the feasibility of function-based screening for detecting hazardous biological functions.
• To propose a framework for implementing function-based screening.

Main Methods:

• Examined the feasibility of function-based screening, focusing initially on proteins.
• Proposed a framework for implementation, starting with toxins and progressing to complex functions.
• Discussed challenges and outlined a research and development strategy.

Main Results:

• Biologically functional proteins are constrained by biophysical and biochemical requirements, making them modelable.
• Function-based screening can detect hazardous sequences irrespective of similarity to known agents.
• A phased implementation framework is proposed, starting with toxins.

Conclusions:

• Function-based screening is a necessary complement to sequence-based methods for biosecurity.
• Modeling biophysical and biochemical constraints is key to feasible function-based screening.
• Further research and development are required to address challenges in function-based screening.