The powerful Feedback control is exploited in thousands of application, anything from flight and engines to oil and gas, traffic, computers and hard drives. We, however, apply real-time feedback control to the world of bacteria.
Feedback control is composed of several components: model of the system, measurements, observers, controllers:
CyberGenome lab tabs
We are control engineers specialized in process engineering
Our vision is to promote the bio-
Our mission is to control the micro world.
- Cells models:
We are working with Cornybacterium glutamicum. Our strains are engineered to receive external commands with sugars via dedicated promoters and to report signal response, e.g. gene expression and growth, pH. Our models are dynamic, consist of differential equations that describe the change of pH, growth, expression and cellular signaling with time. We strive to develop the simplest possible model. We believe simplicity is the key to efficient control.
Control is often worthless without feedback. In self-driving cars for instance, feedback is acquired from several measurements such as cameras, proximity sensors, GPS. This is done real-time as the car is driving. Such feedback measurements, however, are not traditionally exploited in the micro-world. We work to acquire these signals from bacteria cells. We measure expression with lasers (FCM), production and sugars with Liquid Chromatography and use different online NIR probes (OD, cell viability). Since feedback control needs rapid measurements, we automate the whole process of sampling by autosampler and multiplexer (see list of laboratory equipment).
Measurements are always corrupted by noise. This is double true for measurements in micro-world, in which interference is extreme. For example, if we want to measure real-time optical density (OD) of our cells in the culture (useful for growth feedback), we need to account for bubbles (air, oxygen, CO2), and for particles in the medium, for dead cells, etc. The probes usually cannot see all these and may interpret these noise as "real cells". The quality of feedback control is dependent on the quality of the feedback, and we will rather have accurate measurements. We therefore apply advance filtering techniques (usually with statistics) that identify what is the real signal and what is the noise to increase the performance of our controllers.
These are the back and bones of the feedback control: they receive the measurements, and compare to desired reference, for instance our desired production level. The controller then compute the action we need to take (the "control action") that is required to get the response to the desired point. For example in cars, these control actions are steering wheel and the motor thrust. In the micro-world, The reference is a specific growth rate, or expression level, and the control actions can be specific sugar additives such as IPTG, arabinose etc. that change the expression.
The combination of model-measurement-observer-controller is called control theory. Unique in Norway, our laboratory exploits this theory using automatic feedback control action, applied intracellulary in bacteria.