The Qian Lab

California Institute of Technology

How can we rationally design and synthesize
molecular systems with programmable behaviors?

Life is full of amazingly sophisticated programs encoded in genomes, orchestrating molecules to sense, to compute, to respond, and to grow. One approach to interpreting the molecular programs that nature creates is to explore and re-realize the principles of information processing in biology, for example by rationally designing and synthesizing molecular systems that exhibit programmable behaviors. This approach is not only philosophically important, but also practically appealing. As much as electronics has changed our lives with laptops, smart phones and vehicles, at a much smaller scale computer science in its new form as molecular programming will change our lives with nanomachines, smart drugs and diagnostic devices.

The field of DNA computing and molecular programming started with using DNA molecules as a computing substrate to solve hard math problems, where large numbers of molecules were recruited to compute in parallel, presumably much more efficiently and with much less energy than electronics. But the dream of competing with electronics hasn't panned out. Another dream, which has begun to bloom, is to build molecular devices to interact with and manipulate matter at the nanoscale -- the finest scale that you can imagine. For example, a biomolecular computer could be put inside individual cells, collect information from the molecular environment, and regulate the cell’s behavior.

From self-assembly of nanostructures to nanomechanical devices, from molecular finite state automata to nucleic-acid digital logic circuits, researchers are learning to manipulate molecules to carry out designed programs. As a result, these systems have been applied to study DNA-protein interactions, to determine membrane protein structures, to sense pH conditions in living organisms, and to image multiple mRNA expression patterns; such molecular instruments will advance molecular biology in many ways. Artificial nucleic-acid systems have also been applied to organize nanoparticles, to create an assembly line for building molecular objects, and to control the synthesis of novel chemicals; such approaches will have great influence in material science and chemical industry. Artificial nucleic-acid systems have further been applied to engineer biosynthesis pathways, to identify cancer cells, and to deliver antibodies to the targeted cells; such applications will change the width and depth of biotechnology and medicine.

Biochemical circuits

The power and mystery of life is entangled within the information processing at the heart of all cellular machinery. Engineering molecular information processing systems may allow us to tap into that power and elucidate principles that will help us to understand and appreciate the mystery.

DNA is an excellent engineering material for biochemical circuits because its biological nature supports technological applications in vivo, its easy chemical synthesis facilitates practical experiments in vitro, its combinatorial structure provides sufficient sequence design space, and the Watson-Crick complementarity principle enables predictable molecular behavior.

A large variety of DNA circuits have been built with deoxyribozymes or with the help of restriction enzymes. More recently, the introduction of toehold-mediated DNA strand displacement, as a principle for controlling the kinetics of DNA machinery, enabled enzyme-free DNA circuitry that is automated by hybridization alone.

We are interested in designing nucleic-acid strand displacement systems to create ever-more-complex biochemical circuits and molecular devices. We borrow concepts that enabled the success of computer engineering: starting by developing molecular components, trying to understand their behaviors when integrated as systems, then designing these systems with abstractions and programming languages.

Example project: scaling up digital computation with DNA

To construct sophisticated biochemical circuits from scratch, one needs to understand how simple the building blocks can be and how robustly such circuits can scale up. Using a simple DNA reaction mechanism based on a reversible strand displacement process, we experimentally demonstrated several digital logic circuits, culminating in a four-bit square-root circuit that comprises 130 DNA strands. These multilayer circuits include thresholding and catalysis within every logical operation to perform digital signal restoration, which enables fast and reliable function in large circuits with roughly constant switching time and linear signal propagation delays. The design naturally incorporates other crucial elements for large-scale circuitry, such as general debugging tools, parallel circuit preparation, and an abstraction hierarchy supported by an automated circuit compiler.

[1] Scaling up digital circuit computation with DNA strand displacement cascades. Lulu Qian and Erik Winfree. Science, 332:1196-1201, 2011. (pdf) (supp info)
[2] Perspective: "Scaling up DNA computation" by John Reif, 332:1156-1167. (pdf)
[3] News and Views: "DNA computes a square root" by Yaakov Benenson, Nature Nanotechnology, 6:465-467. (pdf)

Example project: spatially-organized biochemical circuits

In biology, nature uses well-mixed circuitry as well as spatially-organized circuitry. Sometimes the latter is much more efficient. In the brain, neurons interact with each other through the physical contact of synapses. In cells, protein scaffolds bring together multiple components of a signaling pathway. Inspired by these biological networks, we seek to create spatially-organized molecular systems that enable faster and more reliable DNA circuits, continuously active molecular devices, and machines with memory.

DNA neural networks

So far, many of the promising examples of designed molecular systems are inspired by computer science. However, one could ask these questions: Besides the most successful approaches of computer science, such as digital logic circuits, are there better ways for engineering molecular computing systems? Could the programming languages that biology uses be completely different, but more efficient and more suitable for engineering molecular systems?

For example, the impressive capabilities of the mammalian brain -- ranging from perception, pattern recognition and memory formation to decision making and motor activity control -- have inspired their re-creation in a wide range of artificial intelligence systems for applications such as face recognition, anomaly detection, medical diagnosis and robotic vehicle control. It is natural to ask: can a soup of interacting molecules act like a network of neurons and exhibit brain-like behaviors?

We are interested in designing and creating molecular systems that exhibit autonomous brain-like behaviors, to embed rudimentary "intelligence" within biochemical systems, capable of learning from their own environment and making decisions in response to a variety of biological signals.

Example project: neural network computation with DNA

Before neuron-based brains evolved, complex biomolecular circuits provided individual cells with the ‘intelligent’ behavior required for survival. However, the study of how molecules can ‘think’ has not yet produced artificial chemical systems that mimic even a single neuron. Building on the richness of DNA computing and strand displacement circuitry, we show how molecular systems can exhibit autonomous brain-like behaviors. We systematically transform arbitrary linear threshold circuits (an artificial neural network model) into DNA strand displacement cascades that function as small neural networks. Our approach even allows us to implement a Hopfield associative memory with four fully connected artificial neurons that, after training in silico, remembers four single-stranded DNA patterns and recalls the most similar one when presented with an incomplete pattern. Our results suggest that DNA strand displacement cascades could be used to endow autonomous chemical systems with the capability of recognizing patterns of molecular events, making decisions and responding to the environment.

[1] Neural network computation with DNA strand displacement cascades. Lulu Qian, Erik Winfree, and Jehoshua Bruck. Nature, 475:368-372, 2011. (pdf) (supp info)
[2] News and Views: "DNA and the brain" by Anne Condon, 475:304-305. (pdf)

Example project: molecular systems that learn

Biological programs have a significant capability unmatched by any man-made computer programs: the capacity to learn. In the brain, a different set of synaptic weights for each neuron corresponds to a different memory, and learning happens when the strength of the receptors in a synapse changes in response to the environmental signals. In our previous work on synthetic DNA neural networks, the learning algorithm was run in silico and the memory was implanted into the networks by pipetting a specific amount of each synaptic weight molecule into the test tube. To take this approach one step further, we seek to create DNA neural networks with weight molecules that can switch from inactive to active, allowing the synaptic weights to be autonomously tuned in response to a variety of biological signals and thus enabling the DNA neural networks to learn from their own environment.

Molecular robots

It is amazing how "dumb" animals such as insects, birds and fish show intelligent behaviors as groups. For example, ants can find the shortest path from the nest to food no matter where the food source is. Termites can gather wood chips into piles no matter how the wood chips are originally distributed. On the other hand, despite wide interest in building molecular robots, the complexity of such demonstrations remains low. A state-of-the-art DNA robot can take a few steps, follow a two-dimensional track, make a choice at a branch, and pick up cargos. To carry out more complex tasks, we are interested in developing molecular robots with collective behaviors, inspired by swarm intelligence in animals. With a simple set of functions such as random walking, modifying tracks, picking up and dropping off cargos, we seek to show that molecular robots can be programmed to perform sophisticated tasks such as solving mazes and sorting targets.