NA-MPNN: Advanced Protein-Nucleic Acid Design Now Integrated into Vecura
This update enables structural biologists and protein engineers to design protein-nucleic acid complexes and predict binding specificity directly through an integrated workflow on Vecura, eliminating the need for complex local infrastructure setup.
What is NA-MPNN?
NA-MPNN (Nucleic Acid-Message Passing Neural Network) is an advanced computational model from the Baker Lab designed to jointly model proteins and nucleic acids (DNA/RNA). By extending the established ProteinMPNN architecture with a unified token alphabet, it allows for simultaneous reasoning across heteropolymer complexes while preserving intricate structural context, such as protein-DNA hydrogen-bond networks. It helps researchers perform structure-conditioned sequence design and predict protein-DNA binding specificity with high structural accuracy. It is especially useful for engineers designing novel DNA-binding proteins or researchers investigating the specificity of transcription factors.
What can users do with NA-MPNN on Vecura?
With NA-MPNN on Vecura, users can:
- Design novel biomolecular sequences: Generate new amino acid, DNA, or RNA sequences for specific chains within a protein–nucleic acid complex.
- Predict protein-DNA binding specificity: Compute position probability matrices (PPMs) to understand which nucleotides a protein preferentially contacts at specific binding sites.
- Refine binding interfaces: Redesign protein-DNA interfaces to optimize or alter binding characteristics.
- Streamline structural analysis: Input complex PDB structures and obtain structurally compatible sequence designs or binding motifs through a simplified, unified interface.
What the output means
The output provides structured FASTA sequences containing combined amino acid and nucleotide codes, as well as position probability matrices (PPMs) formatted as NumPy archives and accessible tables.
This output should be used to support scientific decision making. It does not replace experimental validation.
Why this matters
The ability to accurately model protein-nucleic acid interactions is a fundamental challenge in synthetic biology and drug discovery. Traditional models often focus exclusively on proteins, failing to capture the complex, cross-chain interactions that define how proteins recognize and interact with genetic material.
NA-MPNN bridges this gap by integrating structural information from both partners in a single inference step. This innovation enables researchers to design more effective gene-regulatory proteins and gain deeper insights into the mechanisms of DNA binding, ultimately accelerating the development of customized therapeutic and diagnostic tools.
- Developed by: Baker Lab
- Source: Official GitHub Repository
- Reference: bioRxiv 2025.10.03.679414v2
Vecura で NA-MPNN を試す。
モデルワークスペースを開き、ご自身の入力で評価を始めましょう。