- Slides: 28
Combining Analysis and Synthesis in a Model of a Biological Cell Ken Webb & Tony White SAC ’ 04 March 17, 2004
Introduction In this presentation I will cover: 1. 2. 3. 4. Some background Cell. AK (Cell Assembly Kit) Autopoiesis and SCL (example/test case) Enhanced Cell. AK
Background n Started as exercise in bio-inspiration n n Getting architectural ideas from biology that can be applied to developing complex computer systems But there also aspects of interest to members of cell modeling community n n n “Whole-cell modeling” (M. Tomita, E-CELL) “Grand challenge of modeling multi-cellular animal” (D. Harel) SBML compatible tools
Basic Rationale n In existing cell/biochem modeling tools n n Using OO software development techniques n n Process of building complex cell models is easier for humans (with reuse) Cell. AK: example of such an OO system n n n Each object is a separate act of human design Can have very large number of components Paper accepted by journal Bio. Systems Enhanced Cell. AK: introduced in this paper n Extends Cell. AK to allow it to model autopoiesis
Cell. AK n n An approach to modeling and simulating cells, and other similar biological and non-biological entities. Based on: n n an object-oriented (OO) paradigm, UML visual formalism, ROOM visual formalism. Prototype implemented using Rational Rose Real. Time (RRT)
Cell. AK – some benefits n Scalable, through use of Object instantiation from classes, n Multiplicity, n Chemical metaphor. n n Easy to implement new behavior n If you know C/C++
Cell. AK – Multi-step Process Cell. AK incorporates a top-down process based on current practice in development of embedded and real-time systems. Add more detail at each step. 1. 2. 3. 4. 5. Identify entities, inheritance and containment hierarchies Establish relationships between entities Define entity behavior patterns Implement detailed behavior Validate Entities typically all from the biological domain.
1. UML Inheritance Hierarchy
1. UML Containment Hierarchy
1. ROOM Containment Hierarchy
2. Relationships between entities
3. Behavior between entities
3. The configured system
3. Detailed entity behavior
4. Implement detailed behavior V*S v = ────── Km + S // Irreversible, 1 Substrate, 1 Product, 0 Activator, 0 Inhibitor, 0 Coenzyme case Irr_Sb 1_Pr 1_Ac 0_In 0_Co 0: s = sm->molecule[gene->substrate. Id]. get(); n. Times = enzyme. Level * ((gene->substrate. V * s) / (gene->substrate. K + s)); sm->molecule[gene->substrate. Id]. dec( n. Times ); sm->molecule[gene->product. Id]. inc( n. Times ); break;
Bio. Entity n n Our name for objects in a model of biological cells, or other similar complex reactive system. May consist of any combination of: 1. 2. 3. Behavior Fine-grained structure Other bio. Entities
Bio. Entity Types
Autopoiesis n n n “self-making” All entities in an autopoietic system or network participate in the creation and continual transformation of other entities Based on bottom-up synthesis rather than the top-down analysis of original Cell. AK.
Varela/Mc. Mullin SCL Model 3 types of randomly moving entities: 1. 2. 3. 4. Catalyst Substrate Link also Holes Cell. AK: enzyme Cell. AK: small molecule Cell. AK: lipid Cell. AK: water molecules
SCL-GRO source:  Mc. Mullin, B. , and Gross, D. Towards the Implementation of Evolving Autopoietic Artificial Agents. http: //www. eeng. dcu. ie/~alife/bmcm-ecal-2001/ bmcm-ecal-2001. pdf
Cell. AK version of SCL n n n Cell. AK was unable to model autopoietic systems such as in the SCL model. An enhanced version of Cell. AK adds causal dependency to allow this. In bio. Entities that contain both behavior and fine-grained structure (FGS), the behavior may be at least partly dependent on details of that FGS.
Bio. Entity with Dependency
Generic Bio. Entity Network
SCL Bio. Entity Network
Cell. AK – some limitations n Rigid Structure n n Unable to evolve novel structure that can be incorporated into the running system. Based on proprietary tool (RRT) n But, goal here is to present an approach that can be implemented using a variety of software development languages and tools.
Conclusions n Cell. AK can model biological systems with many thousands of components n n OO/UML top-down decomposition (analysis) Enhanced Cell. AK n n n Adds greater ability for bottom-up synthesis Allows active objects to influence other active objects by effecting their constiuent parts Can model autopoietic systems with lots of interdependencies