Tokyo, Japan – The natural world is a rich source of inspiration for developing sophisticated computational systems; and the opposite is also true, with bioinformatics approaches providing keen insight into biological processes. However, to gain an accurate understanding of the instructions that guide biological systems, input quantities are required. This is not always possible, as these systems are quantified by a variety of attributes.
However, these challenges could be overcome following a study set to be published in Physical Review Letters, in which a team of researchers from the Institute of Industrial Science, The University of Tokyo proposes a novel computational method - one that quantifies information flow in biological systems without needing to measure input signals.
Cells respond to changes in their environment with incredible accuracy and reliability, despite the randomness of external signals. This suggests that cellular signaling systems operate at theoretically optimal levels, which can be tested by measuring actual signaling information flow. This is completed using an approach called mutual information.
"Mutual information is a theoretically grounded metric that is used to connect cellular sensing and signaling to decision-making and control," says lead author of the study, Kento Nakamura. "However, measuring it requires knowledge of both the input and the output of the signaling system, which limits its practical applications."
To address this challenge, researchers used a dual reporter system, which can identify fluctuations in external signals from an upstream pathway – meaning the early events in the chain reaction of communication – without requiring information about its source. The two reporters act like twins: they respond to the same hidden upstream signal while retaining their own independent fluctuations. By comparing their behaviors, the researchers can isolate the component driven by the common source and thereby infer the flow of information from the otherwise unobservable input.
"The results were very exciting," explains Tetsuya J. Kobayashi, senior author. "We found that our method could accurately analyze the information flow leading to bacterial motor output in response to chemical signals, using their cellular motors as natural dual reporters."
In addition to proving that their concept worked in a natural system, the team then compared the measured information flow in a bacterial system with theoretical limits for sensory information. The amount of information used for motor control in the bacteria was consistent with these limits, showing the biological relevance of the analytical approach.
"Our findings suggest that this framework could be used to quantify information flow in a wide range of cellular signaling pathways," reports Nakamura.
A key advantage of this method is that it only requires measurement of information output, not input. This means that it could be applied to a variety of natural systems, such as neural circuits and developmental processes, and possibly even artificial intelligence systems.
