Robust signalling entropy estimation for biological process characterisation.

Abstract

Motivation: Signalling entropy measures the uncertainty or randomness in the signalling pathways of a biological system. It reflects the complexity and variability of protein interactions and can indicate how information is processed within cells. Higher signalling entropy often indicates a more dynamic and adaptive state, whereas lower entropy may imply a more stable and less responsive condition. Estimating signalling entropy has become a valuable method for studying and understanding the complexity of biological processes. This measure has the potential to shed valuable insights into various phenomena, including the mechanisms behind cell fate decisions, drug resistance, and disease progression. To examine the molecular changes within a system, signalling entropy is quantified through the integration of expression measurements and protein interaction networks. Experimental and computational issues, such as false positives and additional noise, can all compromise the accuracy of protein interaction networks. Correction methods can be used to mitigate spurious results, correct for experimental bias, and integrate data from multiple sources. However, to date, the effect of such approaches on entropy calculations, together with the impact of different underlying networks, has yet to be evaluated.

Results: Here, we investigate how the topology of distinct protein interaction networks can alter the entropy calculation. We examine the entropy derived from different protein interaction networks. Additionally, we systematically evaluate different correction strategies, outlining their benefits and drawbacks along with identifying the most effective approaches for specific types of data and biological scenarios. This protocol outlines how to optimize the reliability of entropy calculations and ultimately leads to a deeper comprehension of biological processes and disease mechanisms.