Computational Systems Biology TUM WS 2011/12 Lecture 1: Overview - - PowerPoint PPT Presentation

Computational Systems Biology

TUM WS 2011/12

Lecture 1: Overview

2011-10-20

• Dr. Arthur Dong

Things Good To Know

• Math, Physics, Computer Science (Statistics and Programming)
• Life Sciences (Biochemistry and Molecular Biology)
• Bioinformatics (sequence, structure, system)

 Three pillars roughly equal weight; adjustment possible.  Synergistic collaboration among SEM faculties / schools / universities. 

Appreciate complementary strengths + Acquire complementary ways of thinking.



Technical competence + Ability to ask the right questions.

All About You

What's your background, and why are you here?

• If you study bioinformatics, ...
• For those from physical sciences:

 Where else can you do cutting-edge research that matters so early?!  Risk not taking the time to truly understand biology.

• For those from life sciences:

 How about finishing an experiment in days rather than months?!  Could be a steep learning curve at the beginning.

Course Philosophy and Content

• Bio or informatics? Computational or biology?
• Exciting new field (yeast genome 1996)
• From modeling with DE's to mining patient data – pick your shades of gray!
• Focus on significant questions in biology and medicine.
• Theory/tools as means rather than ends.
• Network-based systems biology.

Course Format and Requirement

• Weekly lecture: cutting-edge research rather than “closed” subject
• Critical reading of seminal/representative papers on discussed topics

 Papers != textbooks != Bible  Look for both strengths and weaknesses  Go beyond – Open questions? Next steps? Apply elsewhere?

• Final presentation (30-min, in groups of 2-3 students)

“Classical” Biology (up to 1950s)

• Anatomy – Organs, tissues, cells
• Mendelian Genetics
• Evolution of species

Then came triumph of reductionism...

“Modern” Biology – Molecular Biology of The Cell

• Cell Biology
• Biochemistry
• Molecular Genetics
• Molecular Evolution

Where does bioinformatics come into the picture? Classically: Protein structure prediction Genomics: Sequence search and comparison Functional genomics and proteomics: Networks and systems

AGAGCATGTTGGCCTGGTCCTTT GCTAGGTACTGTAGAGCAGGTGA GAGAGTGAGGGGGAAGGACTCCA AATTAGACCAGTTCTTAGCCATGA AGCAGAGACTCTGAAGCCAGACT ACCTGGGTCCCAATCTTGGGCTT GGTATTTCCTCGCTGTGTGACTCT GGGTAAGTTACTTAACTTCTCTGT GCCTCAGTTCTCTCAAGTGTAAAG TGACGCTTGTAAAAGTGTCTCCTG CAAAAGAAAGGGCTGCTGGGAGG AGGGGTGTCCCTGGTGTGCACTA AGTACAATATGAGTTTGT … … … MGLSDGEWQLVLNVWGKVEADIP GHGQEVLIRLFKGHPETLEKFDKFK HLKSEDEMKASEDLKKHGATVLTA LGGILKKKGHHEAEIKPLAQSHATK HKIPVKYLEFISECIIQVLQSKHPGD FGADAQGAMNKALELFRKDMASN YKELGFQG Genetic Code

Biology/Chemistry Information

Protein Folding and Structure Prediction Sequence determines structure and structure determines function (roughly!) Challenge: Given target sequence, predict target structure Homology Modelling: Target sequence has a homologous sequence with solved structure

• 1. Align the two sequences (crucial step)
• 2. Put target sequence onto homologous structure and “massage”

Need at least 40% homology Target Template

…QNVERLSLRKNHLTSLPASFKRLSRLQYLDLHNNNFKEIPYILT…

?? Threading:

• Inverse folding problem
• 3D profile and pairwise contact potential
• Difficulty with multi-domain proteins or those with no clear domain structures

What if no close homologous structure?

Ab Initio Prediction Molecular Dynamics Physics – throw in electric charge, solvent etc. and minimize the energy function “Logo” Method Assemble library fragments

• Partial success on small proteins
• In general computationally prohibitive
• How does nature work?

Bradley et al, Science 2005

(Traditional) Tenets of Molecular Biology One gene, one protein, one function (or disease) Protein sequence determines structure, structure determines function

• 2nd. str. pred.

Protein Folding

?

Success!

• Partial success on

small proteins, but dead end?

• How does nature

work? Coiled coils Beta barrels Intrinsic disorders … …

Genomics – Producing the “Parts List” Large-scale sequencing of genomes and the resulting data explosion Sequence Comparison:

• Given a query, find “similar” sequences among tens of millions in databases – fast!
• Align a group of related sequences; identify conserved residues or regions for

structure or function prediction.

• Cluster sequences according to different features.

String comparison algorithms and machine learning (regression, clustering, hidden Markov models, neural networks, etc.)

Functional Genomics and Proteomics

Understanding How Parts Work Individually and Together

• Genome-wide mRNA expression profiling; synthetic lethal screening
• Proteome-wide Yeast-2-hybrid screening and co-AP/MS

Protein-Protein Interactions: Stable Complexes

Transient Protein-Protein Interactions Yeast-two-hybrid PPI reconstitutes TF to turn on reporter gene that enables growth on selective media

H IS 3 D BD B A D O RF HIS 3 D BD B HIS 3 D BD B O RF A D O RF A D

protein-gene interactions protein-protein interactions PROTEOME GENOME & EPIGENOME

Citrate Cycle

METABOLOME Biochemical reactions

Systems Biology in The “Omics” Era

Networks – the central platform of Systems Biology

• Protein-protein interaction networks
• Gene-regulatory networks
• Metabolic pathways
• Graph theory
• Statistical mechanics
• Differential equations

Vertical and Horizontal Data Integration!

From Regular Graphs to Complex Networks

• Favorite graphs
• Cliques and bipartites
• Trees
• Cycles
• Lattices
• Favorite problems
• Euler/Hamil. paths
• Chromaticity
• Graph isomorphism

???

Random Graphs and the Erdös-Rényi model

• Construction
• Start with N nodes (>>1)
• Connect each pair with probability p (<<1)
• Properties
• Node degree k follows Poisson distribution
• Short average path length
• Low clustering coefficient (=p)

Poisson distribution

N = 10 p = 0.2 <k> = 1.8

Real-world Complex Networks

• Communication networks
• Telephone lines
• Internet
• WWW
• Social networks
• Co-authorship
• Actor
• Biological networks
• Gene-regulatory
• Protein-protein interaction
• Metabolic

 Are real-world complex networks really random?  What are the organizing principles behind such networks?  How could such networks have evolved?

Network Parameters and Types

Distinct Topology of Viral and Cellular Systems

• Small power coefficient
• Low local clustering

Viral Networks Exhibit Much Higher Attack Tolerance Than Cellular Networks Yeast KSHV

Big Picture

Stelzl et al, Cell 2005 Rual et al, Nature 2005

Combined viral-host analysis: Need known viral-host interactions! Combining Multiple Systems

The Systems Biology of Pathogen-Host Interactions

Uetz / Dong et al, Science, 2006

Virus Adopts Cellular Features upon “Infection”

Develop new tools for network and system analysis Severe lack of tools, even for a single complex network! Network Alignment Network Docking

• Assess impact of coverage and noise on network topology
• Topology should reflect biology
• Statics should reflect dynamics

Goh et al, PNAS 2007

Systems Biology in Medicine – Disease Networks

Cancer Classification

Set Marker (Leukemia) Network Marker (Breast Cancer)

Looking Ahead

Classical Biology Molecular Biology Systems Biology

Whole Parts Whole

• Understand how life works
• Mechanism (and hopefully) treatment for cancer and other diseases
• Synthetic biology, new materials, new energies

Genome Phenome Computational Systems Biology What questions to ask? What stories to tell?

How To Read A Paper

Focus: Technical details or the big picture? Within the paper:

• What's the whole point, the take-home lesson?
• Why did they do what they did? (historical perspective)
• Any parts problematic and could be improved?
• Expected versus unexpected

Go beyond the paper:

• Observation – Question – Hypothesis – Investigation – Application
• What's the next obvious step?
• Can I apply the same ideas/techniques in other areas?

Turn any question into a project (and possibly a paper)!