Reverse Engineering of Biological Complexity

1
Reverse Engineering of Biological Complexity

• Ms. Abigail Arcilla
• Mr. Vincent Chuaseco
• Mr. John Ruero
• 22 February 2003
• De La Salle University

2
Agenda

• To highlight similarities between biology and
  advanced technologies at the systems-level
  organization despite of having different
  composition, evolution, and development
• Introduce Reverse Engineering methods as applied
  to biological systems

3
Motivation

• To provide a systems-level approach to biological
  systems in order to apply reverse engineering
  methods to unravel hidden complexities of these
  systems.

4
What is a System?

• Recall
• A System is an assembly of parts or modules with
  an underlying hidden complexity that governs its
  structure and the dynamics of its modules.

5
System Properties

• Robustness ability to adapt to different types
  of environment also the ability to withstand
  perturbations in structure without a change in
  identity and function (mutational robustness) for
  short periods bridge between engineering and
  biology

• Fragility cascading failure within a network
  but rare in nature

• Evolvability when a complex system gains new
  information about its environment and passes it
  to its offspring

6
System Properties

• Stability small perturbations do not affect
  solution of stable state condition of a dynamical
  system for long periods

• Degeneracy ability of elements that are
  structurally different to perform the same
  function and to produce different outputs under
  different conditions

7
(No Transcript)
8
Example of System

• USS Enterprise 1701-D
• Propulsion
• Science and Navigation
• Tactical
• Structure

9
(No Transcript)
10
Biological System

• In a biological context
• The large numbers of functionally diverse, and
  frequently multifunctional, sets of components
  interact selectively and nonlinearly to produce
  coherent rather than complex behavior.
• -- Hiroaki Kitano, Computational Systems Biology

11
Example of Biological System

• Gene Regulation in Prokaryotes
• In bacterial genes, a site called promoter is
  where RNA polymerase sticks to start gene
  transcription
• Between promoter and gene, a site called operator
  where another protein, the repressor, sticks
• mRNA production is regulated by these two proteins

12
Gene Regulation in Prokaryotes

• The Lactose Operon in Action

13
Lactose Operon, cont.
14
Biological Complexity

• Biological complexity refers to either
• form and function or
• the sequence that codes for such
• Robust, yet fragile
• Dominated by the evolution of mechanisms to more
  finely tune the robustness/fragility tradeoff.

15
Lifes Complexity Pyramid
16
Properties of Biological Complexity

• Mostly hidden
• Governed by modularity and protocols
• Can be measured
• Types include Structural, Functional,
  Hierarchical, Genomic, Sequential, Physical
  complexity

17
Measures of Biological Complexity

• KCS (Kolmogorov-Chaitin-Solomonoff) complexity
  is directly related to randomness
• Physical complexity based on the entropy of an
  ensemble of sequences how often a certain
  sequence appears with a certain probability
• Joint Density Function characterized in terms of
  entropy and mutual information

18
Measures of Biological Complexity

• Logical Depth measure based on how complexity
  evolves from simpler elements, i.e., fractals and
  automata
• Number of Distinct Parts

19
Biological Complexity

• Leads to Evolution
• the amount of information stored about the
  environment of the organism increases

20
Biological Complexity Example

• O2 Regulation

21
O2 Regulation, cont.

• Avoid deficiency and toxicity of oxygen
• Low-pressure pulmonary circulation receives blood
  from right ventricle and delivers oxygenated
  blood back to left atrium
• High-pressure systemic circulation delivers blood
  to all organs starting from left ventricular
  delivery into the aorta

22
O2 Regulation, cont.

• Regulation orchestrated by multi-level feedback
  control mechanisms
• O2 sensor cells in the carotid signal brain
  respiratory centers
• Blood vessels relax to increase blood supply

23
Biological Complexity Example

• T-cell Dynamics
• T-cells are responsible for cellular immunity
• Equipped with sensors to detect between self
  and non-self gene products

24
T-cell Dynamics, cont.

• Robust since they recall previous exposure to
  antigens to dispose of them more efficiently
  without harming the host
• Complex feedback loops ensure that balance is
  maintained

25
Protocols

• Rules which are used to efficiently organize
  highly structured and complex modular hierarchies
  (modules) to achieve robustness
• Facilitate the addition of new protocols and
  organization
• Good protocol promotes both robustness and
  evolvability
• But, they are a source of fragility

26
Why Protocols?

• To allow interfacing between modules, permitting
  system functions that could not be achieved by
  isolated components
• Simplify modeling, abstraction, and verification
• Allow for independent evolvability of components
  and systems

27
Protocols

• More important than the components/modules
  themselves!
• - Csete and Doyle

28
Protocol Example

• LEGO bricks
• Module the bricks/blocks
• Protocol Snap Connection

29
Protocol Example, IT

• TCP/IP

30
TCP/IP
IP
From Hari Balakrishnan
31
TCP/IP, cont.

• Hides the Network Access Layer from users
• Internet Protocol (IP) provides basic packet
  delivery
• Internet Control Message Protocol (ICMP) uses
  the IP datagram delivery facility to send its
  messages
• Host-to-Host Layer responsible for end-to-end
  data delivery TCP and UDP
• Application Layer examples include FTP, telnet,
  SMTP, HTTP, DNS, SNMP, RIP, NFS

TRANSPARENT TO THE USER!
32
Feedback

• Most powerful protocol for robustness to external
  disturbances and internal component variations in
  complex systems

y
d
F
33
Why Feedback?

• Maintenance of equilibrium homeostasis
• According to Olivier Cinquin and Jacques
  Demongeot,
• Negative feedback has same role of guiding a
  generic (biological) system of variable
  properties toward the correct behaviour, by
  having it constantly correct itself, and thus of
  making the system robust against changes in its
  operating conditions or its internal parameters.
• Positive feedback can be necessary for a system
  to preserve its stability.
• http//www-timc.imag.fr/Olivier.Cinquin/roles_of_
  feedback/roles_of_feedback.html

34
Examples, Feedback
Airplane
Airbag
feedforward
Accelerometer
Sensors
35
Example, Feedback
Thermostat
36
Feedback, Biochemical Pathway
Product
X0
X1
Xi
Xn


Biochemical reactions
Initial substrate
Source Doyle lecture
37
Negative Feedback, Mathematical Basis
G
-

K
Source Doyle lecture
38
What is Reverse Engineering

• The process of analyzing a system to identify the
  its components and their interrelationships and
  create representations of the system in another
  form or at a higher level of abstraction.

Source Chikofsky and Cross
39
Why Reverse Engineering in Biology?

• In biology, to unravel hidden complexities
• Allow the creation of bioinformatics tools and
  databases for prediction, modeling, and
  simulation
• Ex. Detecting homology (similarity) between
  protein sequences
• Predict or verify possible gene pathways to
  refine the design of biological experiments
• Types include time series expression profiles,
  steady-state expression profiles

40
Examples of Reverse Engineering

• Lighting a Match

41
Examples of Reverse Engineering

• Fly swatter

Take these guys out with
42
Example of Reverse Engineering in Biology

• RE of Gene Networks (GN)
• GNs are mostly sparse and large
• Identify underlying network structures to
  discover various gene activities
• One method uses Singular Value Decomposition
  (SVD) and Regression to arrive at solution
• Another method uses Metabolic Control Array (MCA)

43
Predicting Life Processes Reverse Engineering
Living Systems
(storage)
DNA
Transcription
Gene Expression
Translation
Proteomics
Proteins
Biochemical Circuitry
Environment
Metabolomics
Phenotypes (Traits)
44
Methods of Reverse Engineering

• Genetic Algorithms
• Invented by John Holland et al.
• Solutions from one population having a certain
  fitness are used to form a new population, which
  then are used to form new ones

45
Genetic Algorithms, cont.

• Tools and Groups
• GAlib (http//lancet.mit.edu/ga/) a C GA
  library
• GENITOR Group (http//www.cs.colostate.edu/genito
  r/) theory and applications of Genetic
  Algorithms, Evolutionary Computation and Search

46
Methods of Reverse Engineering

• Neural Networks (NN)
• Reverse engineering approach to artificial
  intelligence to understand the human brain and
  cognition in order to develop similar
  implementations
• Conventional neural networks use large arrays of
  processing elements, roughly equivalent to
  neurons, each of which is characterized by an
  activity level which is often a continuous
  variable
• Process involves training NNs to learn network
  topology

47
Neural Networks, cont.

• Advantages are good learning capabilities of
  elements, function holistically, graceful
  degradation, fault tolerant, compute continuous
  valued nonlinear functions
• Computational NN is even used to RE
  nanotechnology

http//www.foresight.org/Conferences/MNT05/Abstrac
ts/Meyeabst.html
48
Neural Networks, cont.

• Tools
• Splice Site Prediction (http//www.fruitfly.org/se
  q_tools/splice.html)
• ProtComp Identification of sub-cellular
  localization of Eukaryotic proteins
  (http//www.softberry.com/protein.html)
• GENESIS (GEneral NEural SImulation System) is a
  general purpose simulation platform to support
  the simulation of neural systems ranging from
  complex models of single neurons to simulations
  of large networks made up of more abstract
  neuronal components (http//www.genesis-sim.org/GE
  NESIS/)

49
Methods of Reverse Engineering

• Bayesian Models
• Time-series probabilistic model used for
  representing stochastic models in biological
  systems
• Uses directed, acyclic graphs
• Observes how a random variable evolves over time
• Algorithms learn at each stage of analysis

50
Bayesian Models, cont.

• Let x1, x2,,xn be nodes with a finite set of
  possible states. A Bayesian network is a
  directed, acyclic graph with nodes x1, x2,,xn
  which encodes dependency relations between them,
  together with families of probability
  distributions associated to each of the nodes

51
Bayesian Models, cont.

• Dependency relations imply that BMs can encode
  deterministic relationships
• very promising for modeling pathways due to the
  flexibility in defining variables, flexibility in
  defining connections between variables, and
  probabilistic nature.

52
Bayesian Models, cont.

• Dynamic Bayesian models such Boolean Network
  Model, linear model of DHaeseleer et al.,
  nonlinear model of Weaver et al., allow for
  integration of stochastic phenomenon which is
  inherent in gene expressions
• The main problem with Bayesian networks arises
  from the difficulty of scoring networks of the
  size necessary to study signal transduction
  pathways.

53
Bayesian Models, cont.

• Tool and Algorithm
• K2 score-based algorithm for learning Bayesian
  networks from data (http//www.kddresearch.org/Gro
  ups/Probabilistic-Reasoning/k2.html)
• Bayesian Network Tools in Java (BNJ) is an
  open-source suite of software tools for research
  and development using graphical models of
  probability (http//bndev.sourceforge.net/)

54
Methods of Reverse Engineering

• Singular Value Decomposition
• From solution of unconstrained linear least
  squares problems, matrix rank estimation, and
  canonical correlation analysis
• Ability to detect and extract weak signals from
  noisy gene expression data
• Ability to recover entire network topology based
  from a number of measurements

55
Singular Value Decomposition, cont.

• Applications image processing and compression,
  immunology, molecular dynamics, small-angle
  scattering, information retrieval
• Implementation by combination with
• Regression Method consists of constructing a set
  of feasible solutions that are consistent with
  the measured data and then we use regression to
  select the sparsest one as the solution

56
Singular Value Decomposition, cont.

• Principal Component Analysis (PCA) proposed for
  pre-processing of data for clustering of genes
  with similar traits
• Tools
• SVDMAN (Wall et al., 2001) detect sampling
  problems when the assays correspond to a sampling
  of an ordinal or continuous variable
• SVDimpute (Troyanskaya et al., 2001) implements
  an SVD-based algorithm for imputing/attributing
  missing values in gene expression data
• Project SVD algorithms at Stanford
  (http//genome-www.stanford.edu/SVD/)

57
Current Status of Reverse Engineering

• Projects/Initiatives
• Virginia Bioinformatics Institute Research (VBI)
  (https//research.vbi.vt.edu/)
• M.I.T. Computational and Systems Biology
  Initiative (CSBI) (http//csbi.mit.edu)

58
Current Status of Reverse Engineering

• Proposal by Dr. Pedro Mendes of VBI genes and
  biochemical networks are tightly interconnected
• simultaneous modeling of both
• Reverse engineer data from transcriptomics and
  proteomics
• https//research.vbi.vt.edu/article/articleview/75