Developmental Biology Concepts on Cyber Entities Evolution

1
Developmental Biology Concepts on Cyber Entities
Evolution

• Kawai Chan
• chankw_at_uci.edu

2
Introduction

• All biological organisms are the product of the
  interplay of genetics, evolution (dynamic) and
  developmental processes (static)
• We can apply these concepts on the Bionet to
  evolve the CEs to increase the behavioral
  diversity

3
Dynamic Diversity Development

• Currently, we only use the dynamic scheme to
  generate CEs behavioral diversity
• Mutation
• Crossover
• The disadvantage of this scheme is the CEs need a
  trial and error period to evolve to be fully
  functional

4
Static Diversity Development

• We would like to shorten the trial and error
  period by implementing the static scheme
• Developmental biology

5
What is Developmental biology?

• Developmental biology is a biological research
  field that explores
• how a single fertilized egg divides to a huge
  number of cells and how they autonomously
  specialize in structure and functionality.
• e.g. A human body consists of 60 trillion cells
  (254 types), each of which has different
  structure and functionality.
• how cells gradually form a complex pattern.
• i.e. how they decide their position in an adult
  body
• e.g. nail at the tip of a finger

6
Important Developmental Processes

• Gene Regulation
• Cell Differentiation
• Cell Lineage
• Cell Induction
• Positional Information by Morphogenetic Gradients
• Cell Division and Cell Death
• Cell Adhesion and Cell Migration

7
Gene Regulation

• Two cells are different if they have different
  subsets of active gene
• The differences between cells are emergent and
  can be controlled
• The activity of gene is regulated by the
  regulatory units
• Regulatory units (cis-regulators)
• Represent a specific DNA region
• Transcription factors (trans-regulators)
• Represent some soluble body fluid to bind with
  cis-regulators

8
Regulatory units

• Regulatory units are like switches for the
  specific cells
• The on and off states of the switches are
  influenced by
• The affinity between cis- and trans-regulators
• The concentration of trans-regulators
• The interactions of all the proteins which are
  necessary the transcription of a gene by
  polymerases
• Auto catalytic regulation of the gene once it is
  activated

9
Cell Differentiation

• To create different cell types
• Two cells are different if they have different
  subsets of active gene in genome
• There are two types of differentiation
• Cell Lineage
• Affected by intracellular factors
• Cell Induction
• Affected by signals from other cells

10
Positional Information

• During development, the positions of cell are
  informed by a concentration gradient of a
  morphogen
• Morphogen is a substance that can influence the
  development of cells
• The morphogen can have different effects on
  different cells depending on
• The affinity with the cis-regulators
• The concentration of the morphogen
• Morphogen are placed in the egg from parents when
  the cell is created

11
Cell division and Cell death

• Different cells have different lifetime, e.g.
  neuron cells are not dividing while blood cells
  do
• These differences depend on the different
  regulatory mechanisms. To stop a cell from
  dividing
• Contact inhibition restricted by space
• Growth factors affected by hormones
• Purposed cell death is common in evolution, e.g.
  to avoid distinction

12
Cell Adhesion

• In neurons, axon and dendrites are connected if
  they have a high enough affinity
• These connections work together as a group, e.g.
  to transmit signals from hand to brain

13
Implementation

• Store a gene as a string of integer for each CE
• Two classes of genes are needed
• Regulatory - switches
• Structural functions

14
Implementation

• The activation of a structural gene depend on the
  regulatory gene adjacent to it. Also, several
  regulatory genes can control one or more
  structural gene

15
Implementation

• The transcription factor is used to calculate the
  affinity with the regulatory genes. When the
  affinity for a cell is higher than a specific
  value, the function of the cell is turned on
• TFs are generated from CEs. TFs should also be
  stored in platforms to distribute the populations
  (e.g. A CE doesnt want to reproduce in a crowded
  platform). This is done by manipulating the
  formulas

16
Formulas
17
Implementation

• Depending on which structural gene is activated,
  one of the following can occur
• A TF is produced to regulate the gene activities
• A cell adhesion molecule (CAM) is produced to
  connect cells to each other, if the affinity is
  high enough
• A receptor is produced to regulate the
  communication between the cells
• A artificial function like cell division, cell
  death, or searching can occur

18
Implementation

• Cell differentiation depends on the signals that
  a CE receives, which may or may not have any
  effect on it depending on the affinity and the
  concentration of the substance
• For Bionet, we will only use the method c, i.e.
  communicating through receptors

19
Implementation

• Positional information (morphogen) is implemented
  with a string of integers as the transcription
  factor. These information is stored in each CE
  and platform. It can diffuse to other CEs
• Both cell division and cell death can be
  implemented as structural genes
• CEs should try to stay close with friends or
  those who provide reliable services

20
Whats next?

• Decide what structural genes (functions) we need?
• Consider including more information from CEs,
  e.g. body executable code, data behavior
  factors and weights
• Consider applying mutation and crossover in the
  gene itself, i.e. to apply dynamic development on
  the static one, this may create a more diverse
  behaviors in a given time

21
Whats next?

• Consider to have some totally static development
  and lifetime control of certain CEs. For
  example, security CE.
• Investigate the difference between goal-setting
  evolution and random evolution.

22
References

• Evolving Morphologies of Simulated 3D Organisms
  Based on Differential Gene Expression Peter
  Eggenberger
• Cell Interactions as a Control Tool of
  Developmental Processes for Evolutionary Robotics
  Peter Eggenberger
• Developmental Process inBionet Evolution
  Simulation - Jun Suzuki

23
The EndThank You