UT assistant professor researches programmable molecules

AddThis

Photo Credit: Victoria Smith | Daily Texan Staff

Computation is often considered a man-made phenomenon, but according to David Soloveichik, it doesn’t have to be. Soloveichik, a UT computer and electrical engineering assistant professor, applies the principles of his field to biological settings to create programmable molecules. 

Soloveichik said future applications of programmable molecules may include chemically-controlled cells that can better carry out chemical processes, smart drugs that detect and kill specific cells and more complex nanotechnology. 

“Biologists generally appreciate computer science as a source of tools, but not as a source of ideas and principles, even though every cell in our body must be able to compute,” Soloveichik said. “Every cell takes input and decides what to do next. The whole process of development is a fundamental algorithm we don’t understand.”

Soloveichik’s work is split into two parts: studying the reaction rules that molecules must obey — called chemical reaction networks — and creating these molecules.

These reaction networks take hours to compute relatively simple tasks, but Soloveichik said the end goal is not speed. Instead, the point is to embed logic into living biological systems.

Soloveichik has worked on one application of reaction networks called a chemical caucus. In this network, the goal is to start with two chemicals and convert the minority into the majority chemical, resulting in only one chemical. 

Georg Seelig, a computer science and electrical engineering assistant professor at the University of Washington who collaborated with Soloveichik to develop the chemical caucus, said this application could be useful in medicine. 

“The kind of logic devices we built could be interesting for creating smarter therapeutics and diagnostics,” Seelig said. “For example, [in a diagnostic test] if some chemical is in the majority, it is cancer. If it’s not in the majority, it’s not cancer. That’s not what we’ve done, but you can imagine it being good for that.”

Additionally, Soloveichik has studied reaction networks that utilize “chemical counting” to output exactly one molecule of some input. According to David Doty, a computer science assistant professor at UC Davis who worked with Soloveichik, this could be the first step to sophisticated computation in chemical reaction networks. 

“The task … is to coordinate and find one molecule the rest identify as special,” Doty said. “There are tasks that get easier if a leader takes charge and coordinates. If there’s two agents that think they are leaders, they will start conflicting with each other.”

Soloveichik said that these reaction networks could theoretically work in living cells but that he currently works on them outside of cells. 

“I’m still at the point where I really want to understand how they work, and I don’t want to put them into cells before I completely understand them,” Soloveichik said. “Cells are very messy, and working outside them gives clarity of thought.”

Doty said that these and more complex chemical reaction networks could be used to model basic logic problems without obvious answers.

“We are trying to get computations to happen in an environment where it’s not obvious how to do [a task] with a regular computer,” Doty said. “For example, a smart drug that activates itself depending on a cellular condition in the body. It’s not so much about the computation being complicated. It’s that it necessarily needs to be done in a chemical context.”

Soloveichik said in the next few years he would like to speed up the reactions, correct error and see the networks embedded into cells, but the long-term future of this field is uncertain. 

“There’s a lot to argue about the future [of programmable molecules],” Soloveichik said. “This is a very early stage, and it’s not clear what it will look like in 20 years, when we actually get good at inserting logic into cells.”

However, Doty has high hopes for the future of programmable molecules.

“Learning how to manipulate matter on the molecular level and learning to program matter to manipulate itself would revolutionize huge areas of science in ways I can’t even guess at,” Doty said.