Evolutionary Directed Computation
Based on DNA Mismatch Repair Systems
The most important thing for science is searching for truth. Through numerous discipline scientists reach the point where they need more powerfull techniques and tools for their tasks. Without new tools some of truths are still uncatchable and they will stay where they are - in the unknown space - usually refered as complexity. Complexity theory can be defined as the search for algorithms used in nature that display common features across many levels of organization. In evolution directed methods a key challenge is that only the small fraction of diversity can be characterized regardless of the efficiency of the screening procedure used. For example, a 500-bp gene implies alternatives, but even the most efficient methods are restricted to alternatives.
Materials and methods:This paper has focus on evolution because it already give us the notion of basic principles for increasing complexity and efficiency. Because evolutionary research moved from dinosaur fossils to microorganisms and genes that are evolving much faster, we can get an insights into new paradigms of evolutionary adaptibility, also called evolvability. New emerging science is experimental evolution which allows to see little deeper than Darwin could at his time. Recent scientists success has indentified that several enzymes were clearly designed to produce mutations (DNA mutases). This was the ultimate validation of thirty years old Radman’s iconoclastic hypothesis of process called SOS-induced error-prone repair. This also has changed the notion of nature’s perfection and the dogma that mutation is merely an unavoidable stochastic event due to the limits of the precision of biological processes. What we have now is that genomic mutations are genetically well controled processes and indentifed enzymes become agents by which evolution can evolve rather than be the passive supstrates of evolution. This new insights in evolutionary concepts (rather then laws) and data, that we can get from experimental evolution, could be make a starting point to transfere analog world of evolution into the digital one. Main purpose will be: to use the concept of self-evolvability to create solutions (by targeted and/or untargeted mutations and recombinations) in reasonable time (and cost) through numerous alternatives when the arrival problem (selective pressure) has never been seen before. Through evolution bacterial populations developed such strategies toward perfection. Its succesfull adaptibility, even in the extremely difficult conditions for survival, relies mostly on homeologes recombinations that allow the exchange of the genetic informations between different species.
This work explaines agents of evolvability as extracted data and concepts from experimental evolution and suggests evolutionary directed computation (EDC-model) based on them.
Results:Recently, few remarkable stories of superrecursive algorithms have been reported. New theory suggested as first: SR-algorithms could produce their result without stopping (e.g.sITM); and second: SR-paradigm allows computation that changes computational procedure. It seems that evolution implicates what superrecursive theory postulates. In the first case, superrecursivity (through EDC-model) could be supported by hypermutable genes (level of particular sequences) and/or by genetic mutators (level of bacterial populations). We can also take into consideration permanent production of antibodies of immune system that constatly improve the solutions, and brain/mind communication system (through Soucek’s model called Quantum-Mind) with its continuos quantizing discrete (CQD) processes of information (from micro to macro level of higher organisms: firefly, katydid, birds, and human).
Second feature of superrecursivity could be supported by mismatch repair system specifically by graded deficiency of LPMR (long-patch mismatch repair) which allows interspecies recombinations and by genomic associations (hitchhiking) with favorable mutations generated by mutator activity. SOS regulon agents are “turn on” and LPMR is “turn off” via multi-step procedure. Processes of these agents lead to instant evolution and completely new genes and sequences that have not been in preexisting library (or database). Agents becomes specialists and through second-order selection rapidly converge to the solution. Adaptation is reachable by specifically nested favorable mutations passing through about seven iterations. Inducible SOS agents are crucial for reducing cost of adaptive function as well as targeted hypermutable sequences.
Implications and applications:a)Computing-Conventional models of recursive and subrecursive algorithms include an extra condition where the algorithm has to stop to give a result. But it is not necessery any more. People are working with displays already, and they can be satisfied when the (printed) results are good enough, even if another, posibly better, result may come in the future. Furthermore, usual notion about solving complex problem, are in the course of higher processing power and capacity. But even when we reach the limit of silicon based techology (some projection said 2012), recursive algorithms will be still limited requesting infinite time for solving complex problem. Instead of “brute force” and exhaustive search through the universum space of all the alternatives, superrecursive algorithms supported by EDC model focus on improving program procedure.
This form of computation could be avaliable before exotic technologies such as molecular DNA or quantum computing. Moreover, human creativity multiplied by this computation will allow human-computer cooperation to an unimaginable degree. This approach opens new perspectives for artificial intelligence methods and, for numerous disciplines and human activities, potential applications appear almost limtless.
b)Biotechnology-After we have taken concepts from several remarkable lessons of evolution, now we can go back to biotechnology and experimental evolution with a new tool. By introducing EDC model we could simulate bacterial mission surveyeing through several layers: generate diversity and extract the desired gene from the library of genes and possibly new genes that code for improved proteins. EDC model can be developed very specifically according to some specific laboratory research and could be used for probabilistic models of phylogenetic trees; for finding interactive proteins; or for searching tiny genetic differences among individuals called single-nucleotide polymophism (SNP’s). It could be expanded to include experimental design. What we are going to get is hypotheses that needs to be tested. EDC assigment is for wide range data/solutions areas where clasical recursive models need infinite time.
Keywords: evolutionary directed computation (EDC); agents of evolvability; SOS-regulon agents(SOS); hypermutable agents (HMS, HRS); hitch-hiking(hh); mutator agents; mismatch repair agents (long-patch,LPMR & short-patch,SPMR); second-order selection.