Ph.D. Student, Indian Institute of Technology, Bombay
Design novel molecular systems capable of performing intelligent tasks
We aim to discover algorithms at the molecular level, that enables living systems to exhibit sophisticated behavior. For example we hope to explain how an artificial cell might carry out complex tasks such as inferring it environment and deciding appropriate action with a soup of chemicals as it only computational resource. We also hope to explore the technological reach of these ideas and find novel applications through the design of smart system of molecules that would do useful in-vivo operations. So far we have designed chemical systems that can solve in-silico, classification tasks such as handwritten digit recognition and inference tasks such as computing the posterior given partial observations. Building a physically realizable chemical reaction system is beset with several technological challenges. While we try to design systems that would operate under these technological constraints, we are also working towards identifying and expanding the class of reaction systems that would be physically realizable. We also hope to make fundamental contributions to machine learning. In particular we hope that chemical reaction networks would prove to be a successful alternative to existing models for machine learning and that some of our in-silico implementations of molecular algorithms would give rise to previously unknown machine learning algorithms.
Abstract: The notion of entropy is shared between statistics and thermodynamics, and is fundamental to both disciplines. This makes statistical problems particularly suitable for reaction network implementations. In this paper we show how to perform a statistical operation known as Information Projection or E projection with stochastic mass-action kinetics. Our scheme encodes desired conditional distributions as the equilibrium distributions of reaction systems. To our knowledge this is a first scheme to exploit the inherent stochasticity of reaction networks for information processing. We apply this to the problem of an artificial cell trying to infer its environment from partial observations.
Pub.: 06 Apr '17, Pinned: 30 Jul '17
Abstract: How smart can a micron-sized bag of chemicals be? How can an artificial or real cell make inferences about its environment? From which kinds of probability distributions can chemical reaction networks sample? We begin tackling these questions by showing four ways in which a stochastic chemical reaction network can implement a Boltzmann machine, a stochastic neural network model that can generate a wide range of probability distributions and compute conditional probabilities. The resulting models, and the associated theorems, provide a road map for constructing chemical reaction networks that exploit their native stochasticity as a computational resource. Finally, to show the potential of our models, we simulate a chemical Boltzmann machine to classify and generate MNIST digits in-silico.
Pub.: 19 Jul '17, Pinned: 30 Jul '17
Abstract: We propose a novel molecular computing scheme for statistical inference. We focus on the much-studied statistical inference problem of computing maximum likelihood estimators for log-linear models. Our scheme takes log-linear models to reaction systems, and the observed data to initial conditions, so that the corresponding equilibrium of each reaction system encodes the corresponding maximum likelihood estimator. The main idea is to exploit the coincidence between thermodynamic entropy and statistical entropy. We map a Maximum Entropy characterization of the maximum likelihood estimator onto a Maximum Entropy characterization of the equilibrium concentrations for the reaction system. This allows for an efficient encoding of the problem, and reveals that reaction networks are superbly suited to statistical inference tasks. Such a scheme may also provide a template to understanding how in vivo biochemical signaling pathways integrate extensive information about their environment and history.
Pub.: 10 Jun '16, Pinned: 30 Jul '17
Abstract: Dynamic DNA nanotechnology often uses toehold-mediated strand displacement for controlling reaction kinetics. Although the dependence of strand displacement kinetics on toehold length has been experimentally characterized and phenomenologically modeled, detailed biophysical understanding has remained elusive. Here, we study strand displacement at multiple levels of detail, using an intuitive model of a random walk on a 1D energy landscape, a secondary structure kinetics model with single base-pair steps and a coarse-grained molecular model that incorporates 3D geometric and steric effects. Further, we experimentally investigate the thermodynamics of three-way branch migration. Two factors explain the dependence of strand displacement kinetics on toehold length: (i) the physical process by which a single step of branch migration occurs is significantly slower than the fraying of a single base pair and (ii) initiating branch migration incurs a thermodynamic penalty, not captured by state-of-the-art nearest neighbor models of DNA, due to the additional overhang it engenders at the junction. Our findings are consistent with previously measured or inferred rates for hybridization, fraying and branch migration, and they provide a biophysical explanation of strand displacement kinetics. Our work paves the way for accurate modeling of strand displacement cascades, which would facilitate the simulation and construction of more complex molecular systems.
Pub.: 11 Sep '13, Pinned: 30 Jul '17