With this project I wanted to simulate gene expression. A long list of A, T, C and Gs is randomly generated, and from these genes are decoded. These genes can be activated/expressed using transcription factors, and can affect other genes in various ways. By setting the initial transcription factors, letting genes be expressed and affect each other, and then reading how common certain of these genes are, images such as these can be generated.

Here, every image represents a DNA. Every pixel is calculated with the x and y-coordinates as input to an instance of this DNA. After running the gene expression for about 10 time steps, a red, green and blue value is extracted.

By adding one more input dimension, you can create gifs!

Using a genetic algorithm, these DNAs can be evaluated, cloned, and mutated to select for certain traits. This means they can be trained to look like predrawn images.

I also experimented with expression output becoming input in the next time step and to neighboring pixels. This creates something more like living, breathing organisms.