Abstract
Zearalenone is a mycotoxin frequently found in cereals and pseudocereals, which has shown male reproductive toxicity and synergistic effects with other mycotoxins. Current detection methods for zearalenone are largely based on chromatography or antibody-based assays, and no de novo designed protein receptors are available as recognition elements for biosensors. In this work, de novo zearalenone-binding proteins have been designed using two competing computational pipelines: (i) Tamarind Bio’s all-atom RFdiffusion-ProteinMPNN-RoseTTAFold workflow and (ii) a NISE-CARPdock pipeline combining scaffold generation, LaserMPNN design, and Boltz-2 folding. Three top candidates (RFdiff-ZB1, NC-ZB1, NC-ZB2) exhibited high structural confidence with global average pLDDT values between 0.97-0.98, docking analysis using GNINA produced docking scores of -5.99 to -9.70. Molecular-dynamics-based minimization yielded affinities of -10.48 to -10.92 kcal/mol with minimized pose RMSDs of 0.26-0.68 Å and CNNscores up to 0.80. These results are comparable to in silico benchmarks from natural zearalenone-binding proteins and mutants. This study presents the first comparison between Tamarind Bio and CARPdock-Boltz-2 workflows for de novo mycotoxin binder design and delivers three high-confidence candidates suitable for biosensing. Future work will include extended MD trajectories, experimental binding validation, and integration into impedance-based or other electrical transduction platforms for real-sample detection.
Supplementary weblinks
Title
GitHub repository
Description
All of the codes used and the files obtained can be found in the github repository.
Actions
View 


![Author ORCID: We display the ORCID iD icon alongside authors names on our website to acknowledge that the ORCiD has been authenticated when entered by the user. To view the users ORCiD record click the icon. [opens in a new tab]](https://www.cambridge.org/engage/assets/public/coe/logo/orcid.png)