All Science is Computer Science!

Vasant Honavar
Artificial Intelligence Research Laboratory
Department of Computer Science
Bioinformatics and Computational Biology Graduate Program
Computational Intelligence, Learning, & Discovery Program
Iowa State University

[email protected]
www.cs.iastate.edu
www.cild.iastate.edu
www.bcb.iastate.edu
Talk given at the meeting of the Iowa Undergraduate Computer Science Consortium. Drake University, October 23, 2004.
Copyright Vasant Honavar, 2004.
Outline

Current state of Computer Science Education

Conceptual impact of Computer Science
All Science is Computer Science!
ISU Bioinformatics Experience
Implications for Computer Science Education
 Computer Science

Computer science is often equated with the

information technologies enabled by it:
Hardware and software for personal computing
Internet and the world-wide web
Electronic commerce
Data mining
Digital libraries
Precision farming
Computer assisted surgery
Workflow management
Databases and information systems
Virtual reality
Smart homes
……………..
 Computer Science Education

Computer Science encompasses a body of knowledge

concerning algorithms, communication, languages,
software and information systems.

Computer science education at present tends to focus on

• Imparting the necessary skills for creating information
technology solutions
and / or
• Teaching the mathematical foundations of computer
science - theory of computation, logic, algorithms, and
complexity
 Computer Science Education

Computer science education at present largely ignores the

conceptual impact of computer science on science,
technology, humanities, society

The result is a computer science curriculum that aims to

train either technicians or mathematicians but not a new
generation of scientists and scholars

This needs to change!

 Conceptual impact of Computer Science

Computer science, broadly defined, is the theory and

practice of representation, processing, and use of
information.
Computer Science offers a powerful paradigm for
modeling complex phenomena such as cognition and
life, and representing, processing, acquiring, and
communicating knowledge that is new in the history
of humanity.
 The road to Computer Science

History of computer science is really a history of

humankind’s attempts to understand nous (the
rational mind) -- intelligence – processes of acquiring,
processing, and using information and knowledge.
Aristotle (384-322 BC) distinguishes matter from form
thereby laying the foundations of representation
Panini (350 BC) develops a formal grammar for Sanskrit
Al Khowarizmi (825) introduces algorithms in his text on
mathematics
Descartes (1556-1650) – Cogito ergo sum!
 The road to Computer Science

Hobbs (1650) suggests that thinking is a rule-based

process analogous to arithmetic
Leibnitz (1646-1716) seeks a general method for
reducing all truths to a kind of calculation
Boole (1815-1864) proposes logic and probability as the
basis of laws of thought
Frege (1848-1925) further develops first order logic
Tarski (1902-1983) introduces a theory of reference for
relating objects in a logic to objects in the world
 The road to Computer Science

Hilbert (1862-1943) presents the decision problem – Is

there an effective procedure for determining whether
or not a given theorem logically follows from a given
set of axioms?
Godel (1906-1978) shows the existence of an effective
procedure to prove any theorem in Frege’s logic and
proves the incompleteness theorem
Turing (1912-1954) invents the Turing Machine to
formalize the notion of an effective procedure
 The road to Computer Science

Church, Kleene, Post, Markov (1930-1950) develop

other models of computation based on alternative
formalizations of effective procedures.
Turing and Church put forth the Church-Turing thesis
that Turing machines are universal computers.
Several special purpose analog and digital computers
are built (including the Atanasoff-Berry Computer)
Von Neumann (1956) works out a detailed design for a
stored program digital computer
Chomsky (1956) develops the Chomsky hierarchy of
languages
 The road to Computer Science

Several digital computers are constructed and universal

languages for programming them are developed –
Lisp, Snobol, Fortran…
Von Neumann, McCulloch, Rashevsky (1940-1956),
investigate the relationship between the brain and the
computer
Von Neumann and Morgenstern develop a formal
framework for rational decision making under
uncertainty
Von Neumann (1956) develops a theory of self-
reproducing automata
 The road to Computer Science

McCarthy, Minsky, Selfridge, Simon, Newell, Uhr et al

(1956) begin to investigate the possibility of artificial
intelligence
Dantzig and Edmunds (1960-62) introduce reduction – a
general transformation from one class of problems to
another
Cobham and Edmunds (1964-65) introduce polynomial
and exponential complexity
Cook and Karp (1971-72) develop the theory of NP-
completeness which helps recognize problems that
are intractable
The rest is… recent history ☺
 Conceptual impact of computer science

The language of computation is the best language we have

so far for describing how information is encoded, stored,
manipulated and used by natural as well as synthetic
systems

Algorithmic or information processing models provide for

biological, cognitive, and social sciences what calculus
provided for classical physics
 Conceptual impact of Computer Science

Computation: Cognition : : Calculus : Physics (Artificial

Intelligence, Cognitive Science)
What are the information requirements of learning?
What is the algorithmic basis of learning?
What is the algorithmic basis of rational decision
making?
Can we automate scientific discovery?
Can we automate creativity?
 Conceptual impact of Computer Science

Computation: Cognitive Science :: Calculus : Physics

Computer science offers fundamentally new ways to
understand cognitive processes –
Perception
Memory and learning
Reasoning and planning
Rational decisions and problem solving
Communication and language
Behavior
 Conceptual impact of Computer Science

Computation: Life : : Calculus : Physics

(Computational Biology, Computational Ecology ..)
How is information acquired, stored, processed, and
used in living systems – in gene expression, protein
folding, protein-protein interaction, reproduction?
How do brains process information?
How do genes and environment determine behavior?
What does the genetic program for fetal development
look like?
 Conceptual impact of Computer Science

Computation: Biology :: Calculus : Physics

Computer science offers fundamentally new ways to
understand biological processes
Reproduction
Development
Molecular function
Gene regulation and expression
Cellular function
Signal transduction
Brain function
Adaptation and Evolution
 Conceptual impact of Computer Science

Computation: Society : : Calculus : Physics

(Computational Economics, Computational
Organization Theory)
What is the informational and algorithmic basis of inter-
agent interaction, communication, and coordination?
Under what conditions can self-interested rational
agents cooperate to achieve a common good?
How do groups and coalitions form?
How do different social organizations (democracies,
economies, etc.) differ in terms of how they process
information?
 Conceptual impact of Computer Science

Computation: Social Sciences :: Calculus : Physics

Computer science offers fundamentally new ways to
understand economic, and organizational, and social
phenomena
Cooperation and competition
Bounded rationality and economic behavior
Community, coalition, and organization formation
Social contract
Organization, interaction, and communication
Rise and fall of cultures e.g., the Anasazi Indians
 Conceptual impact of Computer Science

Computer science offers fundamentally new ways to

model and understand

cognitive, biological, and social processes

through computational or information processing

or algorithmic models
 Algorithms as Theories

We will have a theory of

learning when we have precise information processing
models of learning (computer programs that learn from
experience)
protein folding when we have an algorithm that accepts a
linear sequence of amino acids as input and produces a
description of the 3-dimensional structure of a protein as
output
bounded rationality when we have an algorithm for rational
decision making under limited information, memory or
computation
 Conceptual impact of Computer Science
Pre-Turing
Focus on physical basis of the universe with the
objective of explaining all natural phenomena in terms
of physical processes
Post-Turing
Focus on informational and algorithmic basis of the
universe with the objective of explaining natural
phenomena in terms of processes that manipulate
information
We understand a phenomenon when we can write a
computer program that models it at the desired level
of detail
When theories and explanations take the form of
algorithms, all science becomes computer science!
 All Science is Computer Science!
Is it any surprise then that Computer Science has given
birth to:
Computational molecular biology and bioinformatics
Computational neuroscience and neuro-informatics
Computational developmental biology
Cognitive science
Computational economics
Computational chemistry and chemo-informatics
Computational organization theory
Medical informatics
Agricultural informatics
Geo-informatics
 All Science is Computer Science!

Is it any surprise then that computer scientists are being

hired by departments of
Biological Sciences
Chemistry
Economics
Engineering
Philosophy
Physics
Psychology
Sociology ……….
 Computational Biology

Case Study -- The ISU Bioinformatics and Computational

Biology Program
What is Bioinformatics?
What is Computational Biology?
ISU Bioinformatics and Biology Program
Computer Science BCB Curriculum
Computer Science BCB Research
 Computational Biology

Algorithmic models provide for biological sciences what

calculus provided for classical physics
The language of computation is the best language we
have so far for describing how information is encoded,
stored, manipulated and used by biological systems
Central problem: Given genomic sequences – text in a
language with known alphabet but unknown syntax and
semantics, and some additional clues, discover the
syntax and semantics!
Goal is to develop information processing or
computational models of biological processes (protein
folding, gene regulation, protein-protein interaction)
 Computational Biology

Representative Problems
Inference of tree of life from DNA sequence data
Characterization of protein sequence – structure –
function relationships e.g., discovery of sequence
and structural correlates of protein-protein
interactions
Genetic network inference from gene expression data
Inference of metabolic pathways
Modeling and prediction of cellular processes
Modeling host-pathogen interactions
 Bioinformatics

Transformation of Biology from a data poor science to a data

rich science
• High throughput Data Acquisition
• Processors, Storage, and Communication Technologies
• Algorithms for information processing
In principle, it is possible to gather, store, access, and analyze
large volumes of data (e.g., sequence data, structure data,
expression data)
The focus of bioinformatics is on the design and
implementation of software tools for data driven knowledge
discovery in data rich biological sciences
 Bioinformatics

Leveraging the ability to gather, store, and process large

volumes of data at increasing rates into scientific
advances requires new algorithms and software for
Data description, organization, visualization
Pattern matching, retrieval
Information extraction and integration
Knowledge representation
Data mining and hypothesis generation
Computer assisted collaborative discovery
 Bioinformatics and
Computational Biology

The focus of bioinformatics is on the design and

implementation of algorithmic and systems solutions to
support data driven knowledge discovery in data rich
biological sciences
The emphasis of computational biology is on the
development of information processing or computational
models of biological processes (protein folding, gene
regulation, protein-protein interaction)
Since hypothesis or model construction is generally data
driven, bioinformatics tools are essential for computational
biology
 Bioinformatics and
Computational Biology

Mission of the BCB Graduate Program

To enable students to explore fundamental

research questions by providing training in
biological sciences and computer science,
mathematics, and statistics
 Bioinformatics and
Computational Biology

Program History

1997 Iowa Computational Biology Laboratory

1998 Formal Coursework Begins
1998 Graduate programs offer Areas of Specialization in
Computational Molecular Biology
1998 Began hiring computational biologists (12 to date)
1999 NSF-IGERT training grant
1999 Bioinformatics and Computational Biology Graduate Program
2000 Laurence H. Baker Center for Bioinformatics and Biological
Statistics
2002 BCB Program Review
2004 Board of Regents Review
 Bioinformatics and
Computational Biology

Research areas
Bioinformatics
Functional and structural genomics
Genome evolution
Macromolecular structure and function
Metabolic and regulatory networks
Mathematical biology
Biological statistics
 Bioinformatics And Computational Biology

Program Overview
One of the first Bioinformatics Ph.D. programs in the US
Over 60 Ph.D. students, one of the largest and strongest
Bioinformatics Ph.D. programs in the US
Over 72 BCB faculty in 14 different departments
$30 million in grants between July 1999-July 2004
2 major training grants
(NSF-IGERT, $2.7M; USDA-MGET, $1.7M)
Bioinformatics Summer Institute (NSF-NIH)
 Bioinformatics and
Computational Biology

Interdisciplinary Training
Background & ramp-up courses
Core courses in computational molecular biology and
bioinformatics
Bioethics courses & training sessions
Electives in Statistics, Computer Science, Biology
Faculty Research Seminar
Student Seminar
Computational molecular biology seminars and
symposia
 Bioinformatics and
Computational Biology

Interdisciplinary Research
Research exploration rotations
Joint mentoring -- `Wet’ and ‘dry’ lab research
experiences and mentoring by a major and co-
major professor (from computational and
biological sciences)
International and industrial internships
Representative Computational Biology research

Computational approach
to prediction of protein-
protein interface
residues from protein
sequence
 Representative bioinformatics research

PROSITE,OPROSITE MEROPS,OMEROPS SWISSPROT, OSWISSPROT

Query Decomposition, Answer Composition

M Query S(θ,D,O)

Mapping
Learning Algorithm

O
Ontology O Ontology O’

Infrastructure for collaborative discovery from distributed,

semantically heterogeneous autonomous information sources
 The ISU Bioinformatics and Computational
Biology Experience

It is possible at an institution like ISU

• to train a new generation of biologists --
computational biologists – who are proficient in both
computer science and biology to pursue a
fundamentally new approach to answering basic
research questions in biology
• to train a new generation of computer scientists –
bioinformaticists – to develop new information
technologies for storage, retrieval, and analysis of
diverse types of biological data and knowledge to
facilitate collaborative scientific discovery in biology
 Conceptual impact of Computer Science
Pre-Turing
Focus on physical basis of the universe with the
objective of explaining all natural phenomena in terms
of physical processes
Post-Turing
Focus on informational and algorithmic basis of the
universe with the objective of explaining natural
phenomena in terms of processes that manipulate
information
We understand a phenomenon when we can write a
computer program that models it at the desired level
of detail
When theories and explanations take the form of
algorithms, all science becomes computer science!
 What Next?
Can we replicate the Bioinformatics and Computational
Biology experience at ISU in
• Agriculture – Agricultural Informatics
• Basic Sciences – Computational Chemistry,
Computational Physics, Chemoinformatics,
Computational Neuroscience, Neuroinformatics
• Engineering – Engineering informatics
• Social Sciences – Social Informatics, Computational
Economics

How?
 What should a Computer Scientist know?

Computer science and a subset of:

Biology, Chemistry, and Physics – molecular biology and
neuroscience
Mathematics and Statistics – Logic, information theory,
probability theory, statistics, game theory, decision
theory
Cognitive science – perception, cognition, language,
action
Social sciences – economics, organizational theory,
communication theory ..
Philosophy – epistemology, philosophy of mind,
philosophy of science
What should every literate person know?

Elements of the theory of computation

Algorithms and information processing
Elements of Programming
Information system design and use
Use of computational models in his or her discipline
Implications for Computer Science Education

Computer Science undergraduate and graduate

students need significant exposure to physical,
biological, cognitive, and social sciences!

All undergraduates and graduate students need

significant exposure to computer science – not just
information technology – across the curriculum!