Treatment for a disease like cancer is much different from treatment for a disease like COVID-19: in the COVID case, a drug or vaccine needs to spread throughout your body to be effective; for many cancers, the target is much, much more precise. At the University of Connecticut, Ying Li – a professor of mechanical engineering – is using supercomputing to help create nanomedicines that can exactingly target cancerous cells.
“A lot of medicines involve intravenous injections of drug carriers,” Li explained in an interview with Aaron Dubrow of the Texas Advanced Computing Center (TACC). “We want them to be able to circulate and find the right place at the right time and to release the right amount of drugs to safely protect us. If you make mistakes, there can be terrible side effects.”
Li has been on the cutting edge of nanomedicine for the last decade – which, of course, requires being on the cutting edge of computing techniques. “My research is centered on how to build high-fidelity, high-performance computing platforms to understand the complicated behaviors of these materials and the biological systems down to the nanoscale,” he said. “I’m a 100% computational person, there’s no dirty hands. Because of the size of these particles, this problem is very hard to study using experiments.”
The computer in question, for the most part, is the Frontera supercomputer at TACC). Frontera, which delivers 23.5 Linpack petaflops, placed ninth on the most recent Top500 list. On Frontera, Li typically uses 500-600 processors for a simulation, though at peak, he has sometimes required 9,000 processors operating in parallel.
“Advanced cyberinfrastructure resources, such as Frontera, enable researchers to experiment with novel frameworks and build innovative models that, in this example, help us understand the human circulatory system in a new way,” said Manish Parashar, director of the NSF Office for Advanced Cyberinfrastructure, which is supporting the project.
Currently, only small fractions of existing nanoparticle drugs (predominantly delivered in spherical form factors) make it to their intended targets – which poses a serious problem. “We know that anti-cancer drug molecules are highly toxic,” Li said. “If they don’t go to the right place, they hurt a lot. We can reduce the dosage if we actively guide the delivery.”
So, in some of his most recent research, Li examined how various nanoparticle forms like long, thin “nanoworms” move through the tight geometries of our vascular systems. “It’s like driving on the highway: construction slows down traffic. Drugs are getting carried by individual red blood cells and dragged into narrow regions and getting stuck,” Li said. “The nanoworm moves like a snake. It can swim between red blood cells making it easier to escape tight spots.”
Now, Li is working to incorporate AI – trained using the data from the Frontera simulations – to further streamline the analysis. “The current computational model covers many important processes, but the whole process is so complicated,” Li said. “We’re currently building the training database for the machine learning aspect of our work. … Then, we can pre-train the neural network using the hypothetical data we take from these simulations so they can quickly and efficiently predict the effects.”
To learn more about this research, read TACC’s Aaron Dubrow’s coverage here.