2023.09.22 - Our paper on EGFR targeted dyes is available in the Journal of Nanobiotechnology
Our research group has developed a new Lepatinib/S0456 conjugate (LP-S) near-infrared fluorescent dye that shows promise for improved imaging and detection of metastatic lymph nodes in oral cancer surgery. Currently, surgeons use a standard dye called ICG to try to identify metastatic lymph nodes. However, ICG lacks specificity and can lead to poor results. In contrast, LP-S specifically binds to EGFR receptors enriched on oral cancer cells. In mouse models, enabling superior fluorescence imaging of drainage pathways and metastatic lymph nodes compared to ICG. LP-S provides higher contrast and longer retention in metastatic lymph nodes, allowing them to be clearly differentiated from normal lymph nodes during fluorescence-guided surgery. With further research and clinical testing, LP-S has the potential to significantly advance lymph node mapping and resection in oral cancer patients, leading to more accurate staging and improved treatment outcomes.
2023.08.26 - Our collaboration with IceKredit finished 6th of 872 teams in Baidu's First Global AI Drug R&D Algorithm Competition
Our multiscale deep learned model of compounds tested for inhibitory activity against SARS-CoV-2 considered both local geometric features and longer scale molecular fingerprints to improve accuracy in predicting the activity of unknown compounds. The work conducted here builds upon our recent JCAMD publication and provides another example of the utility of our methodology.
2023.08.08 - Our paper using Bayesian regression to combine docking and deep generative molecular models is published in the Journal of Computer-Aided Molecular Design
Our research group developed a novel hybrid deep learning model that improves the process of computational drug discovery. The model combines a deep generative network for molecular design with a docking score approximation using Bayesian regression. This allows for efficient reinforcement learning to generate new potential drug molecules optimized for binding to the target protein. In testing against the DDR1 kinase, this hybrid approach yielded molecules with significantly higher docking scores compared to similarity-based methods. The model was able to generate chemically diverse compounds, going beyond analogs of known actives. Analysis showed the high scoring molecules made similar interactions to inhibitors and had comparable binding energies, despite their structural differences. Overall, this hybrid deep learning strategy enables more effective sampling of chemical space to discover pharmaceutically relevant compounds for drug development. It reduces data dependence limitations of deep generative networks. With further advancement, this approach could identify promising new inhibitors against novel drug targets.
2023.06.13 - Associate Professor Butch awarded the RSC Horizons Dalton Prize in a collaboration led by NJU-BME's Professor Hui Wei
Nanozymes are artificial catalytic nanomaterials which exhibit behaviour and characteristics similar to enzymes. High-performance nanozymes have numerous potential applications in the biomedical field, such as for the treatment of cancer and inflammation, as well as in wearable devices.
As part of an interdisciplinary team lead by Professor Hui Wei, Professor Butch helped to demonstrate the superior properties of the designed nanozymes compared to natural enzymes, furthering our understanding of what factors control the catalytic activity of nanozymes. |
C-407 & C-408 Modern Engineering Plaza
|
|