Integrated Circuit Models of Bio-Cellular Networks

Abstract

The core idea driving this PhD research project is that there is a finite set of bio-chemical processes that are recurrently found in bio-cellular networks; designing compact, efficient and robust electronic models for such basic reactions can lead to faster development of electronic mimetics for larger bioprocesses. In this thesis analog MOS transistor models for three of such fundamental bio-cellular processes have been proposed. It has been shown that natural analogies exist between bio-cellular reaction parameters and device parameters of an electronic transistor, and exploiting them simpler and faster electronic circuit simulators can be designed for bio-cellular processes. The proposed models use lesser count of transistors than the existing researches, hence leading to faster and cheaper execution of the hardware. To further strengthen the idea of modularization, these basic electronic modules have been cascaded with minor adjustments to form a complete bio-cellular pathway. The three non-linear bio-processes modeled on silicon substrate in this research are receptor-ligand binding reaction, Michaelis Menten reaction and Hill kinetics. The corresponding electronic implementations for the first two have been validated against the deterministic ordinary differential equation (ODE) models of the mentioned bio-processes. A new set of analogies between entities of electronic and biological domains have been established. It has been shown mathematically and through simulations that the gate voltage in an electronic transistor controls the saturation of device current the same way an enzyme or receptor contains the inundation of production rate in a bio-reaction; hence drain voltage stands analogous to substrate or ligand concentrations. Also, the characteristic relationship between device current and the two types of terminal voltages, drain and gate, allows a transistor to be used as a bio-concentration multiplier. Another set of equivalent parameters have been validated; the effect of dissociation constant in a receptor-ligand binding and Michaelis constant in a Michaelis Menten reaction is proportional to the effect of channel length in an electronic transistor therefore no external silicon circuitry is required to model these bio-constants, hence significantly reducing the size of the corresponding electronic models. The behavior of these electronic circuit models compares acceptably with that of respective bio-cellular reaction as reported in the standard literature of bio-chemistry and molecular biology. The circuit sizes have been compared with other existing efforts to model the same non-linear processes in electronic domain, and the proposed circuits have been found to use lesser number of transistors producing the same behaviors satisfactorily. For the third process, the Hill kinetics, a couple of electronics models have been proposed. Concrete mathematical and behavioral validation of Hill process is still under research. The proposed silicon models of the basic bio-processes, receptor-ligand binding reaction and Michaelis Menten reaction, have been cascaded with subtle alterations to implement the electronic design of a very vital pathway found in living cells, the cAMP-dependent pathway. For this pathway a deterministic ordinary differential equation model has been derived from the existing literature of bio-chemistry to show that the whole pathway cascade involves only a couple of bio-processes occurring repeatedly; hence can be designed rapidly and efficiently by combining the respective electronic sub-modules. Producing integrated circuit chip level electronic mimetics of bio-cellular processes can be beneficial in two ways. First, these chips can serve as faster bio-electronic simulators as compared to their computational counterparts. They can also be integrated with bio-systems as bio-sensors and bio-system regulators due to the compactness of size and power efficiency. Secondly, these cytomorphic designs, build on the lines of highly computational intensive yet power and space efficient bio-systems can help improve designs of computational mechanics, and even set new design direction in this domain.