Synthetic neuromorphic computing in living cells.

Area of Science:

• Synthetic Biology
• Computational Neuroscience
• Bioengineering

Background:

• Neuronal network computational properties have inspired computing systems using perceptrons.
• Neuro-inspired models offer potential for reliable, efficient, and reconfigurable synthetic gene circuit design.

Purpose of the Study:

• To introduce the perceptgene, a perceptron computing in the logarithmic domain for efficient artificial neural network implementation in Escherichia coli.
• To demonstrate the utility of perceptgene in creating functional synthetic gene circuits for diverse computational tasks.

Main Methods:

• Development of the perceptgene, a perceptron model adapted for biological computation.
• Modification of perceptgene parameters to encode functions like minimum, maximum, and average of analog inputs.
• Construction of multi-layer perceptgene circuits for complex logic (soft majority, analog-to-digital conversion, ternary switch) and programmable logic gates (OR/AND).

Main Results:

• Successfully engineered perceptgene devices capable of basic analog computations.
• Demonstrated multi-layer perceptgene circuits performing complex functions and programmable logic.
• Showcased AI-driven optimization of perceptgene-based synthetic gene circuits.

Conclusions:

• The perceptgene provides an efficient platform for implementing artificial neural networks in living cells.
• This neuro-inspired approach facilitates the creation of sophisticated and reconfigurable synthetic gene circuits.
• Perceptgene technology enables advanced computational functions within biological systems and opens avenues for AI-guided bioengineering.