Computacin Evolutiva Protemica Un anlisis de representaciones de ubicacin libre basadas en proteomas

1
Computación Evolutiva ProteómicaUn análisis de
representaciones de ubicación libre basadas en
proteomas utilizando el algoritmo genético
proporcional

• Iván Garibay, Ph.D.
• Office of Research and Commercialization
• School of Electrical Engineering and Computer
  Science
• Evolutionary Computation Laboratory
• University of Central Florida
• igaribay_at_mail.ucf.edu
• http//ivan.research.ucf.edu

Version 2.0 110304 104AM
2
Computación Evolutiva (EC)
Rethinking Evolutionary Computation

• Método de computacional inspirado en el concepto
  Darwiniano de evolución por selección natural.
• Es como crianza de caballos de raza uno
  determina quien es el mejor y dirige y controla
  la evolución
• Computadoras hacen posible evolucionar
  estructuras muy rápido horas o días
• Las estructuras que se crian o evolucionan son
• Vectores (para optimización)
• Programas de computadora (control)
• Programas SPICE (circuitos)
• Estructuras Geométricas (antenas)

3
Aplicaciones EC Circuitos

• Koza GP
• 21 reinvenciones
• 2 nuevas patentes

More info http//www.genetic-programming.org/
4
Antenas

• En el espacio 2004
• NASA Ames Research Center
• Hardware Evolutivo
• Funciona, mejor que la que diseñaron grupo de
  expertos en nasa
• No entienden completamente por que funciona

More info http//ic.arc.nasa.gov/projects/esg/res
earch/antenna.htm
5
Problema Complejidad
Introduction

• Necesitamos herramientas para tratar la
  complejidad
• Computación Evolutiva (CE) ha probado ser
  efectiva
• CE afronta limitaciones debido a espacios de
  búsqueda muy grandes y muy complejos
• Convergencia prematura a sub-optimas
• Estancamiento de la búsqueda
• Efectos negativos epistaticos (interferencia
  genética)
• Destrucción de bloques de construcción genética
  muy largos, etc.
• Problema de la complejidad superar limitaciones
  actuales para poder evolucionar estructuras mucho
  mas complejas

6
Aprendiendo de la Naturaleza
Introduction

• Nature evolve strikingly complex organisms in
  response to complex environmental adaptation
  problems with apparent ease
• Localize and extract principles from nature
• Apply them to design better algorithms

Pictures credit Sanjeev Kumar
http//www.cs.ucl.ac.uk/staff/S.Kumar/
7
Representación es critica
Rethinking Evolutionary Computation
Genotype to Phenotype
Genome (DNA) Organisms Computational
Instance of Evolutionary Problem Structure
Solution Bit String Ordering of cities
for TSP 10 01 11 01 (Boston, NY, LA,
Orlando)

• Representacion adecuadamente de el problema es
  crucial.
• Define the space to be explored
• Mapping between possible problem solutions and
  internal representation space

8
Estructuras de Información
Rethinking Evolutionary Computation

• DNA molecule is an information structure
• Store information digitally (chain of
  nucleotides)
• Nucleotide deoxyribose sugar phosphate
  Nitrogenous base
• Nitrogenous bases Adenine, Thymine, Cytosine,
  Guanine
• DNA is an amazingly efficient, highly specialized
  structure for information storage, replication,
  expression and evolution

image credit U.S. Department of Energy Genomes
to Life Program, http//doegenomestolife.org.
9
Estructuras de Función
Rethinking Evolutionary Computation

• Proteins
• Most elementary building blocks of functionality
• Assembled directly from segments of DNA
• Self-assemble into a characteristic
  three-dimensional shape
• Involved in almost every biological process
• Ultimately responsible for all the organisms
  functionalities

image credit U.S. Department of Energy Genomes
to Life Program, http//doegenomestolife.org.
10
Del Genotype al Phenotype
Rethinking Evolutionary Computation

• Classical Genetics linear relation
• Gene ? Phene (visible trait)
• Gene type (hair color gene) ? Phene type (hair
  color)
• Modern Genetics non-linear relation
• Sum (kigenei) cascade metabolic reactions
  (protein-protein, gene- protein, gene-mRNA, and
  others) Environment ? Phene

Genes
mRNA
Proteins
Metabolic Pathways
Visible Traits
Epigenetic Factors
11
Representaciones en la Naturaleza y en EC
Rethinking Evolutionary Computation

• Nature
• Complex genotype to phenotype mapping
• Genes to proteins, proteins interact in complex
  ways to produce biological function and behavior
• Functional structures proteins

• EC
• Usually direct genotype to phenotype
• Each gene represents one characteristic of the
  problem (similar to have one gene for
  intelligence or tallness, clearly not the case)
• No functional structures involved

12
Genomics y Proteomics
Proteomic Approach

• Unique perspective
• Study complete sets of functional building blocks
  that conform an organism (not single gene or
  protein)
• Genomics focus on the study of organism genomes
  complete set of genes
• Proteomics study of organisms proteomes protein
  complement of genome

13
Resultados Intrigantes
Proteomic Approach

• Complexity not correlated with their genome
• Rice genome contain more genes than human genome
  (Goff, 2002)
• Humans and chimpanzees genomes are 98.7
  identical (Fujiyama, 2002)
• Complexity may be correlated with their proteome

14
Representando como la Naturaleza(revisited
lifes complexity pyramid Barabási, 2002)
Proteomic Approach

• Genomics and proteomics provide a better
  understanding at organism level

Emergent Complex function
Self-organization, interaction networks
Basic biological building blocks
15
Una nueva forma de representar
Proteomic Approach
S
Complex Solution Subject to fitness
evaluation (Organism)
Complexity Building Proteins cooperate, compete
and antagonize. Self-organization,
self-assembly (proteome)
Low complexity building blocks encode solution
subject to crossover, mutation, etc. (genome)
Proteins (Functional BBs)
Genes (Information BBs)
16
El Método Proteómico
Proteomic Approach

• Introduce two fundamental departures from
  traditional EC
• The focus of our work is the study of the effects
  of introducing such an interaction space into EC,
  as modeled by a multiset

1. Interaction Space
2. Functional Units
17
Resultados Publicadosen revistas arbitradas

• Journal of Genetic Programming and Evolvable
  Hardware 3(2), pp. 157-192, Kluwer Academic
  Publishers Wu A.S., Garibay I., (2002), The
  Proportional Genetic Algorithm Gene Expression
  in a Genetic Algorithm
• Introducimos el Algoritmo Genético Proporcional
  PGA
• Análisis matemático y estadístico inicial de la
  representación PGA
• Experimentalmente probamos que PGA es tan
  competente o mejor que el GA
• Genoma bloques de construcción muy peculiares

18
Resultados Publicadosen revistas arbitradas

• IEEE Transactions on Systems, Man and Cybernetics
  Part B 34(3), pp. 1423-1434, IEEE Press Wu A.S.,
  Garibay I., (2003), Intelligent Automated
  Control of Life Support Systems Using
  Proportional Representations
• Aplicamos el PGA a un problema muy complejo
  sistema dinámico acoplado
• NASA Sistema de Soporte de Vida para misiones
  largas en el espacio
• Proteínas mejoran resultados de GA

19
Resultados Publicadosen revistas arbitradas

• Journal of Genetic Programming and Evolvable
  Hardware, To Appear, Kluwer Academic Publishers
  Garibay I., Wu A.S., Garibay O.(2006), Emergence
  of Genomic Self-similarity in Location
  Independent Representations favoring positive
  correlation between the form and the quality of
  candidate solutions
• Propiedad clave para el éxito de EC
• Correlación positiva entre la forma y la calidad
  de las soluciones a prueba
• Mostramos experimentalmente que genomas del
  Método Proteómico se auto-organizan en
  estructuras auto-similares
• Probamos formalmente que el Método Proteómico
  favorece esta propiedad clave

20
Otros Resultados Publicadosen conferencias, etc.

• Garibay I., Wu A.S., Garibay O., (2006),
  Emergence of Genomic Self-Similarity in Location
  Independent Representations Favoring Possitive
  Correlations Between the Form and Quality of
  Candidate Solutions, Genetic Programming and
  Evolvable Hardware Journal To Appear, Kluwer
  Academic Publishers.
• Garibay I., Wu A.S., Garibay O. (2005), On
  Favoring Positive Correlations between Form and
  Quality of Candidate Solutions via the Emergence
  of Genomic Self-Similarity, In Proceedings of
  Genetic and Evolutionary Computation Conference -
  GECCO 2005, Washington, DC, USA, June 25-29, ACM
  Press. pp. 1177-1184. Nominated for Best Paper
  Award
• Garibay I.(2004), The Proteomics Approach to
  Evolutionary Computation An Analysis of
  Proteome-based Location Independent
  Representations Based on the Proportional Genetic
  Algorithm short format official format ,
  Doctoral Dissertation, College of Engineering and
  Computer Science, University of Central Florida,
  Orlando, Florida, 2004.
• Garibay I., Wu A.S. (2004), Emergence of Genomic
  Self-similarity in a Proteome-Based
  Representation, In Proceedings of the
  Self-Organization and Development in Artificial
  and Natural Systems (SODANS) 2004, Workshop and
  Tutorial Proceedings Ninth International
  Conference on the Simulation and Synthesis of
  Living Systems (ALIFE IX) Boston, Massachusetts,
  Sep 12 2004, pp. 9-12.
• Garibay I.,Garibay O., Wu A.S. (2004), Effects
  of module encapsulation in repetitively modular
  genotypes on the search space, In Proceedings of
  Genetic and Evolutionary Computation Conference -
  GECCO 2004, Seattle, USA, Jun 26-30 . Vol. 1, pp.
  1125-1137
• Garibay I., Wu A.S. (2004), Emergent white noise
  behavior in location independent
  representations, In Proceedings of the
  Self-organization in Representations for
  Evolutionary Algorithms Workshop - GECCO 2004,
  Seattle, USA, Jun 26-30 . Workshop Proceedings
  CD.
• Garibay I., Wu A.S. (2004), Workshop on
  Self-Organization in Representations for
  Evolutionary Algorithms Building complexity from
  simplicity, In Proceedings of the
  Self-organization in Representations for
  Evolutionary Algorithms Workshop - GECCO 2004,
  Seattle, USA, Jun 26-30 . Workshop Proceedings
  CD.
• Garibay 0.,Garibay I., Wu A.S. (2004), No Free
  Lunch for Module Encapsulation, In Proceedings
  of the Modularity, Regularity and Hierarchy in
  Open-ended Evolutionary Computation Workshop -
  GECCO 2004, Seattle, USA, Jun 26-30. Workshop
  Proceedings CD.
• Wu A.S., Garibay I., (2003), Intelligent
  Automated Control of Life Support Systems Using
  Proportional Representations, IEEE Transactions
  on Systems, Man and Cybernetics Part B 34(3), pp.
  1423-1434, IEEE Press.
• Garibay O.,Garibay I., Wu A.S. (2003), The
  modular genetic algorithm exploiting
  regularities in the problem space, In
  Proceedings of ISCIS 2003 The International
  Symposium on Computer and Information Systems at
  Antalya, TR, Nov 3-5 , LNCS series by
  Springer-Verlag, pp. 578-585.
• Garibay O.,Garibay I., Wu A.S. (2003), The
  modular genetic algorithm motivation and first
  results on repetitive modularity, In Proceedings
  of Genetic and Evolutionary Computation
  Conference Late Breaking Papers - GECCO 2003,
  Chicago, USA, Jul 12-16 , pp. 100-107
• Garibay I., Wu A.S. (2003), Cross-fertilization
  between Proteomics and Computational Synthesis,
  In proceedings of the 2003 AAAI Spring Symposium
  Series---Computational Synthesis at Stanford.
• Wu A.S., Garibay I., (2002), The Proportional
  Genetic Algorithm Gene Expression in a Genetic
  Algorithm, Genetic Programming and Evolvable
  Hardware 3(2), pp. 157-192, Kluwer Academic
  Publishers.
• Wu A.S., Garibay I., (2002), The Proportional
  Genetic Algorithm Representation, In Proceedings
  of Genetic and Evolutionary Computation
  Conference, GECCO 2002, p. 703, Morgan Kaufmann
  Publishers.
• Wu A.S., Garibay I., (2002), The Proportional
  Genetic Algorithm, Proceedings of the Bird of a
  Feather Workshops, Genetic and Evolutionary
  Computation Conference, GECCO 2002, p. 200-205,
  AAAI.
• Garibay I., (2000), Generating Text with a
  Theorem Prover, Proceedings of the 6th Applied
  Natural Language Processing and 1st Meeting of
  the North American Chapter of the Association of
  Computational Linguistics, Student Research
  Workshop, pp. 13-18.