cs4030: biological appications of computing science (biocomputing): introduction & overview...
TRANSCRIPT
CS4030: Biological Appications of Computing Science (BioComputing):Introduction & Overview
George M. Coghill
Structure of CourseStructure of Course
• Lectures – (Wednesday @ 9 & Friday @ 9):– Weds: Taylor A21; Fri: King’s NK14
• Practicals (Friday @ 13:00):– in Room Meston 204, 2 hours per week– Attendance mandatory– Only CS4030 work to be done during this
time: attendance credited only in this case.
Assessment
• 75% from a 2 hour examination in January; the paper will consist of three questions - candidates have a free choice of two from three.
• 25% from continuous assessment
Reading List
• Bioinformatics and Model-based Technology
Recommended:
Krane D E & Raymer M. L. Fundamental Concepts of Bioinformatics. Benjamin
Cummings, 2002. (Library)
May also be consulted:
Kuipers B. J. Qualitative Reasoning, MIT Press, 1994
• Evolutionary Computing
May be consulted:
Mitchell T. Machine Learning (ch 4 & 9)
plus web based material.
AttendanceYou are expected to attend all the lectures. The lecture notes (see below)
cover all the topics in the course, but these notes are concise, and do not contain much in the way of discussion, motivation or examples. The lectures will consist of slides (Powerpoint and possibly OHP transparencies), spoken material, and additional examples given on the blackboard. In order to understand the subject and the reasons for studying the material, you will need to attend the lectures and take notes to supplement lecture slides. This is your responsibility. If there is anything you do not understand during the lectures, then ask, either during or after the lecture. If the lectures are covering the material too quickly, then say so. If there is anything you do not understand in the slides, then ask.
In addition you are expected to supplement the lecture material by reading around the subject; particularly the course text.
What is BioComputing?
• For the purposes of this course:1. The use of computational
methods to solve biological problems (bioinformatics and systems biology).
2. The development of novel compuational methods inspired by biological processes.
Breakdown of the Course
• Bioinformatics:– Including: data searches and pairwise
allignment
• Model-based Technology: – Including: constraint based reasoning and
model learning
• Biologically Inspired Computing:– Including: neural nets, genetic algorithms
and artificial immune systems.
What is Bioinformatics?
ComputationalBiology
Bioinformatics
Genomics
Proteomics
Functionalgenomics
Structuralbioinformatics
What is Bioinformatics?DNA (and RNA) Proteins
Over time, genes accumulate mutations Environmental factors
• Radiation
• Oxidation Mistakes in replication or
repair Deletions, Duplications Insertions Inversions Point mutations
Protein Folding
Why is Bioinformatics Important?
• Applications areas include– Medicine– Pharmaceutical drug design– Toxicology– Molecular evolution– Biosensors– Biomaterials– Biological computing models– DNA & RNA computing
Biologically Inspired Computing
• Neural Nets
• Evolutionary Computing – Genetic Algorithms, Genetics Programming
etc.
• Artificial Immune Systems
• Particle Swarm Optimisation
• Ant Colony Optimisation
xn
x1
x2
Input
(visual input)
Output
(Motor output)
Four-layer networks
Hidden layers
Genetic algorithms
• Variant of local beam search with sexual recombination.
Genetic algorithms
• Variant of local beam search with sexual recombination.
Lecture 1 CBA - Artificial Immune Systems
Multiple layers of the immune system
Phagocyte
Adaptive immune
response
Lymphocytes
Innate immune
response
Biochemical barriers
Skin
Pathogens
Lecture 1 CBA - Artificial Immune Systems
Clonal Selection
QML-CSA: Clonal Selection Algorithm
selection
Antibody repertoire
Selected Antibodies
proliferation
cloned Antibodies
matured Antibodies
AffinityMature
Hyper-mutation
Selected Antibodies
Reselection
Random Antibodies
Update Repertoire
Memorycell
An Evolutionary Algorithm Inspired by the clonal selection principle of immune system
Using hyper-mutation and re-selection instead of crossover and mutation.
Model-based Technology
• Qualitative Reasoning– Symbolic, using no numbers– Structural though incomplete– Synonyms: Naive physics, Qualitative modelling, Qualitative
simulation, Commonsense reasoning, Deep knowledge.
• Developments– Use of any models in the domain reasoning process– Numerical, Interval, Semi-quantitative, Fuzzy, Qualitative,
Rule-based, Procedural
Motivations• Problems with RBS
– Reasoning from First Principles– Dangers with “nearest approximation”
• Second Generation Expert Systems– Use deep knowledge – Provide explanations of reasoning process
• Commonsense reasoning– Capture how humans reason– Enable use of appropriate causality
• Model reuse– Improved ease of ES maintenance
Models and Inference
LearningEngine
Input Data
BehaviourModel
InferenceEngine
Input Data
BehaviourModel
24
Qualitative Modelling• Behavioural Abstraction
PhysicalSystem
ActualBehaviour
DifferentialEquation
Fi: R R
Qualitative Constraints
BehaviouralDescription
numerical or analytic solution
qualitative simulation
25
Qualitative Analysis
??x1↑ f10 ↑
Δx = u – f10
Time
x1 {+,0}
{+,+}
{+,-}
{0,+}
u is steady & positive, how will x and f10
change?
Qualitative Prediction
Quantitative Prediction{+,-}
magnitude
Rate of change
1
u
f10 =k10.x1
€
′ x 1 = u − k10x1
x1’ = u – f10
f10 = M+(x1)
PL models of genetic regulatory networks
• Genetic networks modeled by class of differential equations using step functions to describe regulatory interactions
b
-
B
a
-
A
- -
xa a s-(xa , a2) s-(xb , b1 ) – a xa .
xb b s-(xa , a1) s-(xb , b2 ) – b xb .
x : protein concentration
, : rate constants : threshold concentration
• Differential equation models of regulatory networks are piecewise-linear (PL)
de Jong et al 2003
State transition graph
• Closure of qualitative states and transitions between qualitative states results in
state transition graph
Transition graph contains qualitative equilibrium states and/or cycles
a1 maxa0
maxb
a6
b1
b2
D2 D3 D4
D7
D5
D6
D1
D8 D9 D10
D11 D12 D13 D14 D15
D16 D17 D18
D24
D20
D21 D22 D23
D19
D25
QS3QS2QS1 QS4 QS5
QS10
QS15
QS20
QS25QS24QS23QS22QS21
QS16
QS11
QS6
QS7
QS12
QS17 QS18
QS19
QS13
QS14
QS8 QS9
de Jong et al 2003
Model Learning - compartmental
• Robust to Noise! Learning with Noise
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1
1 3 5 7 9 11 13
No states present
Accuracy
Clean
Inv' N
Rand' N1 2
u
k.x1
k.x2
ko.x2
Glycolysis
30
The Diagnostic ProcessBiologicalSystem (Plant)
Predictor
CandidateGenerator
DiscrepencyDetector
Input Output
Fault Identification
Fault Isolation
Fault Detection
Cascaded Solution Space
x1’=0x2
x1
x2 ’=0
111
12
6 2010
13
7
5
3
9
8
4
1 2
u
k12.x1
k20.x2
8
4
The End?
• A Machine with a Mind of its Own“Ross King wanted a research assistant who would work 24/7
without sleep or food. So he built one.” Wired 12/8/04http://www.wired.com/wired/archive/12.08/robot.html?
pg=2&topic=robot&topic_set=
• The Robot Scientisthttp://www.nature.com/cgi-taf/DynaPage.taf?file=/nature/journal/v427/n6971/abs/nature02236_fs.html&dynoptions=doi1096277730
• Discovery Net
http://www.discovery-on-the.net/