About the Lab
The lab is hosted in the Department of Computer Science and Information Engineering at National Taiwan University, working across computational molecular design, clinical metabolomics, quantum computing, medical image computing, and AI agent systems.
The lab studies molecules and metabolites by computation. Its two founding directions were computational molecular design and biomarker detection in metabolomics; in recent years these have been joined by quantum computing, medical image computing and AI agent systems, giving five research pillars. We develop algorithms and software tools, and apply them to drug development, environmental and regulatory chemistry, and clinical assessment.
In computational molecular design, the lab took part in six drug development projects under the National Research Program for Biopharmaceuticals, working with the chemistry and pharmacy departments of China Medical University, National Tsing Hua University and Taipei Medical University among others. The targets were treatments for neuropathic pain, thrombosis and cancer, and the work holds several U.S. patents. The earlier predictive models cover hERG cardiotoxicity, human oral drug absorption, and QSAR analyses of skin penetration and skin sensitization; the recent ones predict P-glycoprotein substrates with interpretable graph neural networks, curate a CYP450 interaction dataset covering most of phase I drug metabolism, and use deep learning to speed up the non-targeted annotation of per- and polyfluoroalkyl substances.
In metabolomics, the lab built a research group at the Metabolomics Core Laboratory of the NTU Center of Genomic Medicine, developing mass-spectrometry and NMR computing methods, and received the Best Informatics Paper award at an International Conference of the Metabolomics Society. Those methods have been applied to welding-fume exposure, compound identification and toxicity prediction in Chinese herbs, and markers for lung cancer, leukaemia, chronic obstructive pulmonary disease and mortality among intensive-care patients; the recent clinical work includes a human breathomics database and a clinical breathomics dataset, gut microbial networks and metabolic pathways in multiple system atrophy, and the identification of metabolites in children with attention-deficit hyperactivity disorder.
The three newer directions each have published results: quantum computing for drug discovery and patent search, including a comparison of two quantum programming platforms for quantum chemistry; medical image computing, which grew out of the lab’s gait work into an open-source distributed platform for 2D video-based movement analysis and now reaches clinical image analysis and translation; and AI agents and generative models for the automatic assessment of negative symptoms in schizophrenia.
Besides papers and public tools, the lab has run GreenN (SAS), the Searching, Assessment and Screening System for Safer Alternative Chemicals, since September 2021, commissioned by the Chemicals Administration of the Ministry of Environment. It helps companies assess the hazard and risk of a chemical, how closely it is regulated, and what it can be replaced with, supporting early-stage R&D decisions and regulatory compliance; the project is ongoing. "AI-Powered Green Chemistry Diagnostic Advisor", built on that framework, received a Team Excellence Award at the Presidential Hackathon 2025 and the Special Award in the 2026 NTU Electrical Engineering Class of 1975 Alumni Endowed Award for Technology Research Innovation.
Since the lab was founded in 2006 it has published 113 journal articles, 44 of them since 2020, together with 98 conference papers and 45 patents. We maintain eight public services under cmdm.tw -- among them 3Omics, Lipidpedia, CypRules, GC2MS, Chromaligner and PITracer -- all open to external researchers. The lab currently has 38 members and 68 alumni.