A cellular automaton featuring structural variants in cells.

• create the grid of cells.
• Colour every mutation with a random colour.
• Mutation probability sample from a distribution.
• Model cell data as n bases rather than genomic regions.
  • Mutation on a 'x' number of bases chosen at random.
• Impl Distribution for SvState in order to randomly sample a Sv State for mutation.
• Colors as a function of the total state of the genome.
  • Model 3 broad type of mutations as the 3 channels of colour { gain: red, loss: blue, inv: green}
  • With every mutation, slowly decrease the corresponding channel by the difference over NATURAL_SELECTION and prop of genome affected till it reaches black
• Pause functionality.
• Local natural selection.
• Model germline mutations as border colour --> can be used to track lineages.
• Model somatic mutations as the fill colour --> to check heterogeneity.
• Model the entire grid as a tissue.
• Find nearby cells (6 nearby cells).
• Differing strength of natural selection as a function of selection strength of nearby cells.
• Show generation as a label somewhere.

1. Each cell starts with no mutations.
2. Each cell has a genome which is a flattened vector, where each bin is represented by n_sv times. The vector has 0 and 1s only.
3. Cells cycle through at constant times.
4. Some cells will randomly have lower natural selection threshold for the duration of one S phase. This is environmental perturbation and will start things off for mutations to snowball.
   1. Will make this interactive at later stages.

1. At every G1 phase, it picks a neighbour number at random [0, 7].
   1. Boundary checking needs to be done for the cells at the boundary, which will have lower number of neighbours.
2. Then it does a hamming distance calculation between that neighbour and itself. Implemented using XOR between the flattened vectors
3. The hamming distance is converted to natural selection by (1 - dist). So the motivation is the cell is trying to be like one of its neighbours.

1. Mutation probability is sampled from a beta distribution.
2. A single bin ID is chosen at random. One bin mutation per S phase
3. The bin is mutated with a SV type chosen at random.
   1. Mutation happens only when the bin is reference. If it is already mutated, nothing is done.