Abstract
Biotoxins, primarily produced by venomous animals, plants, and microorganisms, exhibit high physiological activity and unique therapeutic effects, such as blood pressure regulation and analgesia. While several venom-derived drugs are on the market, drug design resources—particularly extensive activity data—remain insufficient. To address this, we developed the Biotoxins Database (BioTD), the largest open-source repository for toxins. BioTD provides access to 14,607 data records (including 8,185 activity records) covering 8,975 toxins from over 900 species. Activity data is categorized into five groups: Activity, Safety, Kinetics, Hemolysis, and other indicators. BioTD also includes data on 986 mutants, refined sequence information, and disulfide bond annotations. BioTD is freely accessible at http://biotoxin.net/.
Introduction
Nature has spent millions of years optimizing biotoxins—peptides and proteins from venomous organisms—to target physiological systems with extreme precision. These toxins have already yielded blockbuster drugs like Captopril for hypertension and Ziconotide for pain management. However, the transition from "venom to pharmacy" is frequently hindered by fragmented data. While general resources like UniProt or the PDB provide sequences and structures, they often lack the specific, experimentally validated activity data (e.g., $IC_{50}$ values or lethal doses) essential for drug design. BioTD bridges this gap by centralizing sequence, structure, and functional activity into a single, searchable platform.
The BioTD Platform: Methods and Implementation
The BioTD database was constructed through a combination of automated data mining and rigorous manual curation. The authors performed systematic literature reviews across PubMed and patent databases, distilling data from 5,220 references.
Key technical features include:
- Sequence Refinement: Using the NovoPro tool, the team distinguished between "whole sequences" and "signal peptide sequences," allowing researchers to identify the mature, active peptide.
- Structural Modeling: For the ~78% of toxins without experimentally resolved structures, the platform integrates high-confidence 3D models predicted by AlphaFold.
- Web Implementation: Built on a PHP/MySQL backbone, BioTD integrates BLASTp for sequence similarity searches, enabling users to find related toxins via FASTA input.
Database Composition and Scale
BioTD integrates and cross-links data from established benchmarks, including UniProt, PDB, AlphaFold DB, DrugBank, and NCBI Taxonomy. The resulting resource is currently the largest of its kind.
The database contains 14,607 data records covering 8,975 toxins and 678 targets. Notably, 100% of the toxins have visual 3D structures, and over 56% of the records contain specific activity data.
Key Insights and SAR Support
A primary focus of BioTD is supporting Structure-Activity Relationship (SAR) analysis. Unlike general protein databases, BioTD specifically catalogs 986 mutants. This allows researchers to observe how single amino acid substitutions affect binding affinity or toxicity, a critical step in lead optimization.
Furthermore, the data highlights the continued importance of ion channels in toxin research. Potassium ($K_v$) and Sodium ($Na_v$) channel toxins account for 35% of the target-related records. Activity data is meticulously categorized into five groups: Activity (effective dose), Safety (lethality), Kinetics ($K_d, K_i$), Hemolysis, and Other physiological indicators.
Conclusion and Future Directions
BioTD represents a significant advancement in biotoxin research, providing a comprehensive toolkit for venom-derived drug discovery. The platform is freely available at http://biotoxin.net/ and features interactive 3D visualization. Future work will focus on continuous data mining and the integration of AI-driven tools for molecular generation and virtual screening, evolving BioTD from a repository into an active drug discovery engine.
Summary updated