What if biochemical systems could learn similar to neural networks? Our new study explores how dimerization networks “evolve” the ability to classify inputs—chemical style.
Collaborators: Alexei Tkachenko (Brookhaven National Laboratory), Bartolo Mognetti (Université libre de Bruxelles) and yours truly.
The ideas about dimerization networks recently developed by Jacob Parres-Gold, Michael Elowitz and collaborators go back to my papers on mass action equilibrium in protein interaction networks:
-
S. Maslov, I. Ispolatov
Propagation of large concentration changes in reversible protein-binding networks
PNAS 104(34): 13655-13660 (2007).
PNAS link
Reveals how concentration perturbations transmit (or not) through networks via mass action. -
S. Maslov, I. Ispolatov
Spreading out of perturbations in reversible reaction networks
New Journal of Physics 9, 273 (2007).
NJP link
Introduces a mathematical framework for fluctuation propagation in mass-action-governed networks and maps them to current flow in resistor networks. -
K.K. Yan, D. Walker, S. Maslov
Fluctuations in Mass-Action Equilibrium of Protein Binding Networks
Physical Review Letters 101, 268102 (2008).
PRL link
Focuses on fluctuations in mass-action equilibria. -
J. Zhang, S. Maslov, E.I. Shakhnovich
Constraints imposed by non-functional protein-protein interactions on gene expression and proteome size
Molecular Systems Biology 4:210 (2008).
MSB link
Non-specific protein interactions restrict gene expression and proteome size. -
J. Zhang, S. Maslov, E.I. Shakhnovich
Topology of protein interaction network shapes protein abundances and strengths of their functional and nonspecific interactions
PNAS 107(52): 22599-22604 (2010).
PNAS link
Network topology sets protein abundances and governs functional vs. nonspecific interactions.