If the 20th century was the "Age of Physics", then without doubt the 21st century will be the "Age of Biology". This revolution in biomedical research has been driven by the introduction of high-throughput experimental technologies that generate enormous data sets. The ensuing "data deluge" opens up exciting new opportunities to computer scientists because the scientific potential of the data can be exploited only if they are analyzed with sophisticated computational methodologies. In the not too distant future the majority of biomedical researchers are expected to be data analysts rather than experimental biologists or chemists. This is why pharmaceutical companies and biotech enterprises already offer very attractive career paths to computer scientists with strong experience in bioinformatics, data mining/biostatistics, AI/machine learning and related fields.
The main aim of the "Introduction to Computational Biology" (ICB) course is to teach Computer Science majors how to apply their mathematical and computational knowledge to practical data analysis problems in the pharma and biotech industries. Only high-school level biology is required; we rather focus on the essential "soft skill" of how to communicate with biologists. In short, the course provides the intellectual foundations for those students who may later decide to join the "biotech revolution", contributing to new discoveries of great relevance to society.
Synergies with other courses:
The ICB course is strongly recommended to those taking the "Data mining", "Structure and dynamics of complex networks" or the "Deep learning" courses as these methodologies are immediately applicable to biological data analysis tasks.
The course consists of two parts: Biocybernetics and High-throughput data analysis. Biocybernetics (A. Aszódi) addresses information processing in living systems, in particular the analysis of gene expression regulation and metabolic control. High-throughput data analysis (P. Sarkozy) focuses on the analysis of large-scale genomic sequencing experiments, with special emphasis on medical decision support.
Motivation and perspective:
You will gain specific skills in how to understand and model biological phenomena, and learn how to collaborate efficiently with biologists and chemists working in pharmaceutical research. In addition, you will learn scientific methodologies applicable to model complex phenomena in any discipline, further enhancing your employability.
- Good mathematical skills. We will teach you how to use basic calculus (differential equations), linear algebra (vectors, matrices) and basic probability theory.
- Python 3 programming skills.
- High school biology. We will teach you all the necessary knowledge beyond that level.
- W.R. Ashby: An Introduction to Cybernetics
- Klipp, E. et al: Systems Biology. Wiley, 2016.
- P. Prusinkiewicz, A. Lindenmayer: The Algorithmic Beauty of Plants
Topics in detail
Part I: Biocybernetics (A. Aszódi)
Introduction to biocybernetics:
Definition of systems. Comparison of natural and artificial systems. Applicability of systems theory and engineering in biology. Components of biological systems. Biological time series. Modelling biological phenomena.
Stock-and-flow model of inflation. First-order chemical reactions and reaction networks. Systems of linear ordinary differential equations. Solving linear ODE systems by finding the eigenvalues/eigenvectors of the coefficient matrix. Qualitative behaviour of linear systems.
Bulk-phase mass action kinetics. Reversible reactions, equilibrium. Isomerisation catalysis. Enzyme kinetics models: steady-state approximation, Michaelis-Menten rate equation. Competitive and noncompetitive inhibition.
Principles of genomic regulation:
The molecules of life: DNA, RNAs, proteins. The "central dogma": transcription and translation. The differences between prokaryotic and eukaryotic organisms. Prokaryotic genome regulation: the E. coli lac operon, the lambda phage bistable switch, the "Repressilator" synthetic genetic network. Eukaryotic genome organisation: chromosome structure, splicing, epigenetic regulation. The self-regulatory Hes1 network.
Oscillations and chaos:
Periodic phenomena in living systems. The Byelousov-Zhabotinsky chemical oscillator. Biological examples of oscillatory phenomena: glycolysis, mitotic oscillations, predator-prey interactions (Lotka-Volterra models). Chaotic dynamics. The Lorenz attractor and the logistic map. Chaos in enzyme-catalysed reactions. Qualitative behaviour of nonlinear differential equations, the linearisation approach. Detecting chaotic behaviour with Lyapunov exponents.
Stochastic biochemical kinetics:
Basic probability theory. Bayes' Rule. Principles of stochastic kinetics. Master equation approach, the Gillespie stochastic simulation algorithm. Properties of Markov chains. "Convergence" of stochastic kinetic trajectories to the bulk kinetics model in the limit of very large number of molecules.
How to construct mechanistic hypotheses from observations. Popper's falsifiability theory. Kinetic indistinguishability. System analysis by structural or parameter perturbation. Robustness of genetic networks. Self-regulation in biochemical networks: metabolic control theory, flux balance analysis.
The theory of evolution:
Fundamental concepts. Lamarckian and Darwinian evolution. Evolution of macromolecular sequences, molecular phylogeny. Epigenetic inheritance.
Computing with biomolecules:
Adleman's DNA-based solution of the travelling salesman problem. Molecular implementations of Boolean logic gates. Computing with enzymatic reaction networks. Simple learning phenomena.
Scalar product of functions. Orthogonal decomposition. Polynomial and trigonometric bases. The Fourier series and the Fourier transform. Signal processing in the inner ear. Solving differential equations with the Fourier transformation. Power spectral density, Wiener-Khinchin theorem. Rate constants from noise in chemical equilibrium: the Feher-Weissman method.
Regulation in spacetime:
Algorithmic models of plant growth, applications in computer graphics. Turing's theory of morphogenesis. Modelling reaction-diffusion networks with partial differential equations. Solving the diffusion equation with Fourier transformation techniques. The Gierer-Meinhardt model of pattern formation in Dictyostelium discoideum. Robustness of pattern formation in living systems.
Chemoton theory: self-reproducing autocatalytic reaction networks. Cellular automata, Conway's "The Game of Life". In silico models of simulated evolution: the Tierra and Avida systems.
Part II: High-throughput data analysis (P. Sarkozy)
Introduction to molecular genetics:
Role and characteristics of DNA in organisms. Mutation types, population genetics, linkage disequilibrium, transcription and translation of DNA to proteins, gene expression, epigenetic modifications, the Human Genome Project and the path to personalized medicine.
Overview of DNA sequencing technologies:
Sanger sequencing to single-molecule real-time DNA sequencing, in vitro diagnostics, high-throughput measurement methods, partial genetic association studies, genome-wide association studies, single-molecule real-time DNA sequencing.
Quality control, filtering, common failure modes and platform-specific error profiles of common measurement methods, sample multiplexing and study design.
Strings in bioinformatics:
Naïve exact matching, Z algorithm, naïve approximate matching, radix sorting, suffix indices, longest common prefix, Burrows-Wheeler transformation.
Mapping and assembly of large, complex genomes:
Alignment scoring schemes, de-novo assembly, reference mapping, the Smith-Waterman algorithm, the Needleman-Wunsch algorithm, understanding and correcting alignment bias in DNA sequencing, local and global alignment, paired-end sequencing.
Phylogenetics and metagenomics:
Multiple sequence alignment, clustering approaches, distance metric, phylogenetic tree construction, metagenomic population studies, molecular clock hypothesis.
Interpretation of results:
Identifying variants, detecting somatic mutations, heterogeneous population sequencing, construction of local phylogenetic trees for cancer evolution, resolving haplotypes, copy number variations, large-scale genomic rearrangements.
DNA repair mechanisms, homology directed repair, DNA editing in-vivo and in-vitro, CRISPR-CAS9 system, zinc finger nuclease technology, in vivo delivery methods, RNA interference, system biological approach to diseases.
Models in Computational Genomics:
Markov chains, Gaussian mixture models, hidden Markov models, support vector machines, biologically inspired artificial neural networks, neural network training, convolutional neural networks, feedforward-backpropagation algorithm.
Bayesian frameworks in bioinformatics:
Frequentist vs. Bayesian approaches, naïve Bayes classifiers, Bayesian networks, probabilistic classifiers, network structure learning, semantic technologies for computational biology.
The co-evolution of man and machine:
Brain-computer interfaces, the challenges of biosynthetic organisms, the extension of mankind with weak artificial intelligence, challenges posed by strong artificial intelligence.
Data analysis and simulation tasks (reaction kinetics of biochemical pathways, predator-prey models, biological pattern formation models etc.) using the CoCalc on-line computer algebra system.
Data analysis of an in-silico genetic association study using various open-source software packages. Simulation of a DNA sequencing experiment, analysis of the results.
The students prepare an essay (no less than 12000 characters in length without spaces) on biological information processing from a topic list provided by the lecturer. Those who propose a topic on their own will get an extra half grade (i.e. B+ instead of B). Instead of writing an essay it is also possible to write small programs to simulate biological regulation phenomena.
The students write a research report that summarizes their homework assignments and provide an objective overview of their results in no less than 12000 characters in length (without spaces). An additional grade (e.g. B to A, B+ to A+ ) will be given for the use of publicly available real measurement data instead of simulated data.
Grading will be based on the following criteria:-
- Essay and research report (70%): the students must demonstrate that they have understood the principles discussed in the lectures and can apply their knowledge in a practical context. Originality and a critical approach is especially important.
- Course activity (20%): students are required to ask questions and challenge the lecturer and each other.
- Homework assignments (10%): timely completion of the tasks with correct results is required.
András Aszódi (born 1964) studied chemistry at Eötvös Loránd University in Budapest where he graduated in 1988. He then studied molecular neurobiology at the University of Oxford, supported by a Soros scholarship. He received his Ph.D. in 1991 on the kinetic models of simple learning processes. From 1992 to 1996 he developed protein structure prediction methods at the National Institute for Medical Research in London. In 1996 he joined the Novartis Research Institute in Vienna as a computational modeller. He built up the In Silico Sciences unit that provided bioinformatics and computational chemistry tools to researchers. In 2006 he joined the Research Institute of Molecular Pathology in Vienna where he was developing data analysis tools and databases for high-throughput sequencing projects. He is currently teaching scientific programming and biostatistics to PhD students and postdocs. He has over 35 scientific publications, including a book with W.R. Taylor on protein structure prediction.
Peter Sarkozy (born 1984) received his degree in Computer Science from the Budapest University of Technology and Economics in 2009, and continued his graduate studies at the Department of Measurement and Information systems. During his graduate studies from 2009 to 2012 he participated in multiple projects together with the Department of Genetics, Cell and Immunobiology at the Semmelweis University. His areas of interest include the measurement and error characteristics of next-generation DNA sequencing technologies. He is the first person in Hungary to apply Oxford Nanopore Technologies’ single molecule real-time sequencing technology. He is currently working as a research assistant at the Department of Measurement and Information Systems at BUTE.