Building production-grade protein structure prediction pipelines that outperform AlphaFold3. CASP top-ranked · 470+ citations · 30,000+ active users — currently accelerating cancer vaccine design at InstaDeep/BioNTech.
2023 – Present
CurrentDesigned and developed a protein structure prediction pipeline for structural immunology, integrating TCR-pMHC modelling using AlphaFold2, Boltz-2, and Chai-1 to support BioNTech's personalised cancer vaccine programme.
Sep 2020 – Present
ResearchCo-led the development and maintenance of IntFOLD integrated server for 3D protein model prediction, serving 30,000+ users and 400,000+ jobs every year.
Dynamic Ensembles
Capturing conformational diversity beyond static snapshots.
De Novo Design
Modelling novel folds where evolutionary history is silent.
Blending physics-based methods with modern ML to de-risk translational biology projects.
Computational structural biologist with 9+ years of research experience in protein structure prediction, molecular dynamics simulation, and AI/ML-driven drug design. Developer of multiple internationally benchmarked tools (IntFOLD, ReFOLD, FunFOLD, MultiFOLD, ModFOLDdock2Q) that consistently rank among the top methods at CASP blind prediction experiments. Published 14 peer-reviewed papers (H-index 9, 470+ citations) in journals including Nucleic Acids Research, Blood, and Bioinformatics Advances. Currently combining deep learning structure predictors (AlphaFold2/3, Boltz-2, Chai-1) with physics-based methods for cancer vaccine design in collaboration with InstaDeep/BioNTech. Experienced in Python, PyTorch, HPC/SLURM environments, and end-to-end bioinformatics pipeline development.
Iterative refinement
Blend physics-based models with ML to close the loop from prediction to validation.
Operational pipelines
Ship pipelines teams can run, monitor, and extend across structure prediction, docking, and MD.
Collaboration-first
Work closely with domain experts so the science stays rigorous while tooling stays approachable.
"Proteins are not statues. They are breathing, moving molecular machines. I am building tools to capture the dynamic ensembles that drive biology."
Conventional AI treats proteins as rigid shapes. My protocols integrate Molecular Dynamics to reveal how evolutionarily silent targets truly behave.
Using Local Quality Estimation, we lock certain regions while allowing binding sites to breathe, uncovering cryptic pockets for drug discovery.
"We are moving from predicting a single static snapshot to simulating the full movie of life. This shift is critical for engineering better enzymes and designing smarter drugs for traditionally undruggable targets."
Read Full Research ManifestoNext-generation tools for structure prediction, refinement, and quality assessment.
Refinement of 3D protein models using molecular dynamics simulations and quality assessment.
Quality assessment of protein-protein complex models using deep learning and structural features.
Template-based ligand binding site prediction using structural alignment and quality assessment.
Integrated web resource for high-performance protein structure and function prediction.
Peer-reviewed contributions to protein structure prediction, refinement, and quality assessment.
Adiyaman, R., Edmunds, N. S., Genc, A. G., Alharbi, S. M. A., & McGuffin, L. J.
Adiyaman, R., & McGuffin, L. J.
Adiyaman, R., & McGuffin, L. J.
Taylor, K. A., Elgheznawy, A., Adiyaman, R., et al.
McGuffin, L. J., Alhaddad, S. N., Behzadi, B., Edmunds, N. S., Genc, A. G., & Adiyaman, R.
Fadini, A., Adiyaman, R., Alhaddad, S. N., et al.
McGuffin, L. J., Edmunds, N. S., Genc, A. G., Alharbi, S. M. A., Salehe, B. R., & Adiyaman, R.
Edmunds, N. S., Alharbi, S. M. A., Genc, A. G., Adiyaman, R., & McGuffin, L. J.
McGuffin, L. J., Adiyaman, R., Maghrabi, A. H. A., et al.
McGuffin, L. J., Aldowsari, F. M. F., Alharbi, S. M. A., & Adiyaman, R.
Sahli, K. A., Flora, G. D., ... Adiyaman, R., et al.
Kryshtafovych, A., ... Adiyaman, R. (CASP-COVID participants), et al.
Yucesan, G., Yilmaz, A., & Adiyaman, R.
Thesis: 'The Molecular Dynamics Based Refinement of Predicted 3D Models Using Different Restraint Strategies'. Developed ReFOLD2 (top-10 CASP13) and ReFOLD3 (top-5 CASP13, best server CASP14). 4 first-author papers. Supervisor: Prof. Liam McGuffin.
Thesis: 'In Silico Optimisation of Zinc Binding Proteins for Biosensor Applications'. Computational protein engineering using molecular docking and mutagenesis. GPA 3.86/4.00.
Dissertation: 'GaAs/GaAlAs Quantum Well'. Research on semiconductor quantum well structures.
Coursework in classical mechanics, relativity, and quantum mechanics. Foundation for computational physics and simulation methods.
InstaDeep / BioNTech • 2023 – Present
End-to-end TCR-pMHC structure prediction pipeline supporting BioNTech's personalised cancer vaccine programme. Integrates AlphaFold2, Boltz-2, and Chai-1 with a hybrid scoring system (FoldX, Rosetta, DockQ, PatchDockQ, ilDDT) for candidate ranking.
University of Reading • 2020 – Present
Co-led IntFOLD server (30,000+ users, 400,000+ jobs/yr). MultiFOLD3 outperformed AlphaFold3 in quaternary structure prediction by integrating Boltz-2, Chai-1, Protenix and geometric-based rescoring in CAMEO.
University of Reading • 2023 – 2024
Led CASP16 submission achieving 2nd place in protein-ligand pose ranking. Surpasses AlphaFold3 server method in protein-ligand interaction prediction via blind + template docking and affinity scoring.
University of Reading • 2022 – Present
Quality assessment system for protein binders engineered with 16,000+ features combining physics-based and ML scores (DockQ, QS, PatchDockQ, PatchQS, ilDDT, aa-local-lddt, ics). Contributed to CASP15/16 EMA rankings.
University of Reading • 2020 – 2023
MD-based refinement protocol that improves 3D predictions beyond AlphaFold2 accuracy. Integrates local quality estimation with MD recycling; published in Bioinformatics Advances 2023.
University of Reading • 2017 – 2021
Top-10 (CASP13) and top-5 (CASP13) refinement methods; best server at CASP14. MD restraint strategies lifted predicted structures toward native and improved ligand-site fidelity. 4 first-author papers.
Five surprising truths about refinement, assembly, and validation in the post-AlphaFold era.
Why we need to move beyond static snapshots to capture the full range of conformational dynamics.
Integrating deep learning with molecular dynamics and local quality estimation to model mutants and novel folds.
How dynamic modelling will accelerate therapeutic discovery and the design of next-generation enzymes.
We will publish discussions with experts in protein engineering, AI in drug discovery, and computational biology. Follow the channel for upcoming releases.
Auto-synced from YouTube
Follow my research journey and insights on X and Bluesky
Visualizing complex biological systems through computational modeling and design.
Designing novel protein folds with high stability. I use deep learning to hallucinate new backbones and sequence design to stabilize them, creating proteins that don't exist in nature.
Interactive mutational analysis. Select a variant below to see how specific residue changes impact local packing and stability scores in real-time.
Real-time scoring of antibody-antigen interfaces. I optimize sequences to maximize DockQ (structural quality), QS (quaternary structure), and ICS (interface contact score) metrics.
Join my research network across these platforms for daily updates on protein design, structural biology toolsets, and academic insights.
Professional networking, industry news, and formal collaborations.
Connect Now →Fast-paced research signals, library drops, and bio-ML threads.
Follow Updates →Open research notes, daily signals, and lightweight threads.
Join the Feed →Open-source tools, CI/CD pipelines, and protein design codebases.
View Repos →Models, datasets, and community demos for protein design ML.
Explore Hub →Containerized tools, reproducible pipelines, and ready-to-run builds.
Browse Images →Molecular dynamics demos, research walk-throughs, and structural highlights.
Watch Demos →Citations, peer-reviewed publications, and structural biology impact metrics.
View Papers →For deep-dive inquiries, partnership offers, or specific research requests.
Send Email →I am always open to research collaborations, consulting on protein design pipelines, and partnering on structural biology products. Whether you have a question or a project in mind, reach out.
Email me directly
recepadiyaman2244@gmail.com