Postdoctoral Researcher & AI Technical Director

Weining Lin

Postdoctoral researcher at UCL and AI Technical Director at Toursun Synbio, specializing in AI-driven protein engineering, computational biology, and bioinformatics.

Weining Lin
01. Research

Featured Research

Developing AI-powered protein engineering solutions for drug discovery and enzyme optimization.

Research Overview

My research focuses on developing AI-driven methods for protein engineering, including variant pathogenicity prediction, enzyme optimization, and protein function annotation. I work with protein language models, graph neural networks, and generative models to address challenges in drug discovery and synthetic biology. My work spans from fundamental bioinformatics to commercial applications in antibody design and enzyme engineering.

VariPred: AI-Driven Prediction of Missense Variant Pathogenicity

Nature Scientific Reports (2024) | ISMB 2023 Protein Language Models Clinical Genomics Deep Learning

Developed VariPred using transfer learning with protein language models to predict missense variant pathogenicity. Achieved 17% AUC-ROC and 20% MCC improvement over state-of-the-art methods on ClinVar benchmark, eliminating manual feature engineering and providing robust predictions even with missing data.

GOBeacon: Ensemble AI for Protein Function Prediction

Protein Science (2025) | ISMB 2024 Multi-modal Learning Graph Neural Networks Protein Function

Developed GOBeacon, an ensemble model integrating sequence embeddings (ESM2), structure embeddings (ProstT5), and protein-protein interaction graphs via GNN. Applied contrastive learning to achieve state-of-the-art performance on CAFA3 and PDBch benchmarks.

ProFam: Family-Conditional Protein Language Model

EurIPS 2025 (NeurIPS-endorsed) Generative Models Protein Design Fitness Prediction

Created TED database with 370M domains from AlphaFold DB. Developed ProFam, a conditional autoregressive model pretrained on multi-sequence alignments and structural tokens. Outperforms state-of-the-art in protein design and fitness prediction; 30 designs under experimental validation.

AI-Guided Optimization of PETase for Plastic Degradation

PhD Research Project (Oct 2023 - Present) Enzyme Engineering Generative Models Sustainability

Addressing plastic waste crisis through enzymatic biocatalysis. Trained scoring models (SCC=0.93) on 200+ IsPETase variants. Used RosettaFold-All-Atom and LigandMPNN to design 20 novel proteins with 90% scoring higher than experimental variants. Developing diffusion and Bayesian flow models for next-generation optimization.

More publications available on Google Scholar

02. Education

Academic Background

From physiology and neuroscience to pioneering research in AI-driven protein engineering and computational biology.

Ph.D. in Computational Biology and Bioinformatics • University College London

AI and Proteomics Track

Ph.D. in Computational Biology and Bioinformatics • 2022 - August 2025

Supervisors: Prof. Christine Orengo, Prof. Andrew Martin. Research focus on AI-driven variant pathogenicity prediction, enzyme optimization, and protein function annotation using protein language models and deep learning.

M.Sc. in Translational Neuroscience • University College London

Neuroscience

M.Sc. in Translational Neuroscience • 2020 - 2021

Awarded placement on the Dean's List for top-performing students.

B.Sc. in Physiology and Pharmacology • University College London

Life Sciences

B.Sc. in Physiology and Pharmacology • 2017 - 2020

03. Experience

Professional Journey

From algorithm engineering to leading AI-driven protein design initiatives in academia and industry.

AI Technical Director • Toursun Synbio

AI Technical Director

October 2025 - Present

Lead client-facing protein engineering programs including antibody discovery, hit-to-lead progression, and lead optimization. Direct CDR-centric affinity and developability optimization using structure- and sequence-based design. Establish end-to-end pipelines for epitope mapping, binding prediction, and humanization. Lead development of antibody de novo design diffusion models and enzyme engineering R&D.

Postdoctoral Researcher • University College London (Orengo Lab)

Postdoctoral Researcher

October 2025 - Present

Develop full-atom models for protein-ligand binding prediction using LIGSIS and FAMPNN. Train multi-task models for avian influenza mutation effect prediction. Expand PETase variant datasets and design Flow-Matching-based generative models for enzyme optimization.

Applied Computational Scientist Intern • InstaDeep

Applied Computational Scientist Intern

June 2025 - September 2025

Fine-tuned AlphaLink2 on curated antibody-antigen structures. Performed large-scale SabDab data curation with strategic train/test splits. Trained Uni-Fold models with crosslink restraints. Detected epitopes using Chai-1, Boltz-2, AlphaFold-Multimer, and MaSIF.

Research Assistant • University College London (Orengo Lab)

Research Assistant

January 2023 - June 2025

Discovered novel PET-degrading enzymes from MGnify database. Designed scoring systems for PETase-like enzyme evaluation. Applied generative models (LigandMPNN, RFDiffusion-All-Atom, RosettaFold-All-Atom) for enzyme optimization.

Algorithm Engineer Intern • Syneron Technology Co., Ltd.

Algorithm Engineer Intern

April 2023 - August 2023

Optimized ProteinMPNN with MIF-st embeddings for KRAS peptide inhibitor design. Evaluated deep-learning docking models to construct filtering pipeline. Developed GNN model with Masif-based surface features for affinity prediction.

Algorithm Engineer Intern • VecX Biomedicine Co., Ltd.

Algorithm Engineer Intern

January 2023 - March 2023

Developed deep-learning model for AAV-4 variant fitness prediction achieving 0.96 correlation coefficient. Performed structural analysis to identify critical residues for capsid assembly. Integrated fitness predictor into ML-guided AAV engineering workflow.

Current Focus

Currently serving dual roles as Postdoctoral Researcher at UCL (Orengo Lab) and AI Technical Director at Toursun Synbio. My work spans from fundamental research in protein engineering to commercial applications in antibody discovery and enzyme optimization. I bridge cutting-edge AI research with real-world therapeutic development, focusing on protein language models, generative design, and structure-based optimization.

04. Recognition

Awards & Achievements

Recognition for research excellence and contributions to computational biology and bioinformatics.

Dean's List for Top-Performing Students

University College London, MRes Translational Neuroscience • 2021

EurIPS Conference (NeurIPS-endorsed)

ProFam: Autoregressive protein family language model • 2025

ISMB Presentations

GOBeacon (2024) and VariPred (2023) • 2023-2024

05. Connect

Get in Touch

Feel free to connect through these platforms.