Here’s a cool application:
https://barabasi.com/f/1101.pdfWhere they get the initial result
This gives theoretical (without developmental/spatial constraints) gap junction connections between neurons given matrix X with m neurons possibly embodying n gene expression profiles, while matrix O says which gene expression profiles will form gap junctions with one another. B is an m by m matrix which shows which neurons will connect with one another (theoretically). The elements of B/X/O can be boolean, probabilities, or probability distributions. They have to go beyond this model in the paper because of developmental/spatial constraints but I thought it was cool
