Electrostatics, engineered

Charged protein
sequence predictions

We use deep learning to predict and generate protein sequences for highly charged protein-protein interfaces. Not just to fold, but to bind, with precision, flexibility and functional intent.

Explore the research Get in touch

The barnase-barstar complex: two proteins docked together, with the charged residues at their shared interface picked out in amber and cyan. — Barnase bound tobarstar, PDB 1BRS, drawn as a surface with the fold showing through.Amber andcyan mark the charged residues that hold the interface together.Drag to turn.

The research

Designing proteins
nature hasn't built yet

Our latest project focuses on making these designs smarter and more reliable, especially in challenging environments like the bloodstream or the brain. We are improving how our system understands shape, charge and flexibility, so that the proteins we build behave the way we need them to. Not just in theory, but in life.

From gene regulators to next-generation therapeutics, we are creating protein interfaces evolution missed, but modern life needs.

Shape
Complementary geometry across the interface
Charge
Electrostatic match, residue by residue
Flexibility
Behaviour that holds outside the model

Rendering of a protein structure with its electrostatic surface highlighted, showing charged patches across the binding face. — Charged patches across a binding face are what determine whether two proteins meet, and whether they hold.

Where this goes

How our models help the world

We aim to accelerate global research by providing an open-source, fine-tuned model that assists with structural biology, drug discovery and synthetic biology research worldwide.

Targeted cancer therapies
Design custom antibodies for high-charge tumour interfaces
We are developing models that will help scientists generate novel protein sequences tailored for therapeutic binding, to advance precision immunotherapies including cancer vaccines and checkpoint inhibitors.
Next-generation genetic medicine
Create nucleic acid-binding proteins for gene editing
By designing sequences with charge-aware specificity, we hope to enhance CRISPR tools and synthetic regulators for safer, more effective treatment of genetic diseases.
Eco remediation
Engineer protein complexes to neutralise pollutants
Our model aims to help design interfaces for stable binding in extreme conditions, enabling proteins that degrade toxins, bind heavy metals, or break down plastic waste.

Who is building it

Team

Machine learning, molecular biology and delivery, in one group. Select a name to read more.

Tamara DinneenCore project team
Tamara Dinneen leads design of machine learning pipelines that merge technical architecture with real-world operational relevance. She is currently completing a Master's in Machine Learning & AI and has 20+ years experience leading digital transformation, AI, and partner strategy at Oracle, SAP, and Telstra. Her work spans technical AI development, channel analytics, and commercial design of enterprise-scale ML workflows, including those focused on near-real-time video compliance, customer churn prediction, and enzyme sequence design.
Arun SriramanCore project team
Arun is an accomplished AI scientist and innovation leader with over a decade of experience across machine learning, NLP, and applied AI in life sciences and enterprise systems. Now Associate Director at Novartis, he leads AI initiatives in global drug development. At Ion the Fold, Arun spearheads deep learning model design for sequence prediction of charged protein-protein interfaces, applying state-of-the-art techniques from his prior roles at Salesforce, AstraZeneca, and Bank of America. He holds multiple patents and brings rare fluency across both enterprise AI systems and experimental biology models.
Neetu SasidharanCore project team
With over 16 years' experience managing complex IT transformations across government and enterprise sectors, Neetu brings deep expertise in cloud migration, regulatory governance, and enterprise-scale program delivery. Formerly leading digital programs at Schlumberger and the Abu Dhabi Investment Office, she has overseen Oracle and Google Cloud deployments, ISO 27001 alignment, and AI-driven reporting initiatives. At Ion the Fold, she ensures technical execution aligns with rigorous compliance and delivery standards, scaling our research infrastructure while safeguarding integrity and interoperability.
Bernd Willems PhDMentor: molecular biology and genetics
Bernd is a PhD molecular biologist with deep expertise in skeletal development and active-site microenvironments, having published multiple papers on osteogenesis and Wnt signalling. He currently leads synthetic biology technical consulting at Twist Bioscience across APAC, where he supports academic and industrial groups in translating computational designs into DNA-synthesizable constructs. With over a decade of experience spanning R&D, business development, and translational assay platforms (AYOXXA, Qiagen), Bernd brings a unique cross-functional lens to this project, mentoring on both the biological foundations of charged and catalytic systems and the practical constraints of protein design, synthesis, and stability under real-world conditions.
Mohammed Sameer SyedMentor: AI engineering
Mohammed is an AI engineer and ML researcher pursuing his Master's at the University of Arizona. His work bridges generative AI for semiconductor design, digital phenotyping, and LLM systems engineering, from scratch pretraining and LoRA fine-tuning to multi-agent orchestration and scalable model pipelines. He has architected AI platforms, productionised ML and deep learning models, and built free, full-stack tools including end-to-end medical and language assistants. At Ion the Fold he works on protein folding, building interpretable, high-throughput AI systems to decode the language of life.