NTRS - NASA Technical Reports Server

Back to Results
Evolutionary Cell Computing: From Protocells to Self-Organized ComputingOn the path from inanimate to animate matter, a key step was the self-organization of molecules into protocells - the earliest ancestors of contemporary cells. Studies of the properties of protocells and the mechanisms by which they maintained themselves and reproduced are an important part of astrobiology. These studies also have the potential to greatly impact research in nanotechnology and computer science. Previous studies of protocells have focussed on self-replication. In these systems, Darwinian evolution occurs through a series of small alterations to functional molecules whose identities are stored. Protocells, however, may have been incapable of such storage. We hypothesize that under such conditions, the replication of functions and their interrelationships, rather than the precise identities of the functional molecules, is sufficient for survival and evolution. This process is called non-genomic evolution. Recent breakthroughs in experimental protein chemistry have opened the gates for experimental tests of non-genomic evolution. On the basis of these achievements, we have developed a stochastic model for examining the evolutionary potential of non-genomic systems. In this model, the formation and destruction (hydrolysis) of bonds joining amino acids in proteins occur through catalyzed, albeit possibly inefficient, pathways. Each protein can act as a substrate for polymerization or hydrolysis, or as a catalyst of these chemical reactions. When a protein is hydrolyzed to form two new proteins, or two proteins are joined into a single protein, the catalytic abilities of the product proteins are related to the catalytic abilities of the reactants. We will demonstrate that the catalytic capabilities of such a system can increase. Its evolutionary potential is dependent upon the competition between the formation of bond-forming and bond-cutting catalysts. The degree to which hydrolysis preferentially affects bonds in less efficient, and therefore less well-ordered, peptides is also critical to evolution of a non-genomic system. Based on these results, a new computational object called a "molnet" is defined. Like a neural network, it is formed of interconnected units that send "signals" to each other. Like molecules, neural networks have a specific function once their structure is defined. The difference between a molnet and traditional neural networks, is that input to molnets is not simply passed along and processed from input to output units, but rather it is utilized to form and break connections(bonds), and thus to form new structures. Molnets represent a powerful tool that can be used to understand the conditions under which chemical systems can form large molecules, such as proteins, and display ever more complex functions. This has direct applications, for example to the design of smart,synthetic fabrics. Additional information is contained in the original.
Document ID
Document Type
Conference Paper
Colombano, Silvano (NASA Ames Research Center Moffett Field, CA United States)
New, Michael H. (NASA Ames Research Center Moffett Field, CA United States)
Pohorille, Andrew (NASA Ames Research Center Moffett Field, CA United States)
Scargle, Jeffrey (NASA Ames Research Center Moffett Field, CA United States)
Stassinopoulos, Dimitris (NASA Ames Research Center Moffett Field, CA United States)
Pearson, Mark (NASA Ames Research Center Moffett Field, CA United States)
Warren, James (NASA Ames Research Center Moffett Field, CA United States)
Date Acquired
August 19, 2013
Publication Date
February 1, 2000
Subject Category
Computer Programming and Software
Distribution Limits
Work of the US Gov. Public Use Permitted.

Related Records

IDRelationTitle20000064579Analytic PrimaryWelcome to the NASA High Performance Computing and Communications Computational Aerosciences (CAS) Workshop 2000
Document Inquiry