|
|
| Lecturers: | prof. dr. ir. Marcel Reinders, Delft University of Technology dr. ir. Dick de Ridder, Delft University of Technology dr. K. Anton Feenstra, Vrije Universiteit, Amsterdam drs. Nicola Bonzanni, Vrije Universiteit, Amsterdam dr. Aalt-Jan van Dijk, Wageningen University & Research Center dr. Gunnar Klau, CWI, Amsterdam |
| Contact: | dr. ir. Dick de Ridder e-mail: d.deridder@tudelft.nl telephone: +31 15 2785114 |
The next course will be given January 16-20 2012, at the Faculty of Electrical Engineering, Mathematics and Computer Science, Delft University of Technology, Mekelweg 4, Delft, The Netherlands (Building 36). Travel directions can be found here. Most lectures will take place in the Snijderzaal, on the first floor. If you come in, take a left after the "Servicepunt", another left and go up the stairs.
The course is aimed at PhD students with a background in bioinformatics, computer science or a related field; a working knowledge of basic statistics and linear algebra is assumed. The NBIC Technology Track course "Pattern recognition" and the ASCI course "Advanced pattern recognition" (a1) discuss many of the tools used in this course, but it is not required to have followed these. Prior knowledge of molecular biology is a bonus, but also not strictly required.
Preparation material on probability theory, linear algebra and molecular biology can be found below and should be read by all students before the course starts.
Molecular biology is concerned with the study of the presence of and interactions between molecules, at the cellular and sub-cellular level. In bioinformatics and systems biology, algorithms and tools are developed to model these interactions, with various goals: predicting yet unobserved interactions, assigning functions to yet unknown molecules through their relations with known molecules; predicting certain phenotypes such as diseases; or just to build up biological knowledge in a structured way.
Such interaction models are often best modelled as networks or graphs, which opens up the possibility of using a large number of readily available algorithms for inferring networks, performing simulations of biology, optimising paths or flows through networks, graph-based data integration and graph mining. Many of these algorithms can be applied (sometimes with slight alterations) to solve a particular biological problem, such as modeling transcriptional regulation or predicting protein interaction/complex formation, but also to derive systems behaviour by breaking down networks into modules or motifs with certain characteristics.
In this course, we will first give a brief overview of molecular biology, the advent of high-throughput measurement techniques and large databases containing biological knowledge, and the importance of networks to model all this. We will highlight a number of peculiar features of biological networks. Next, a number of basic network models (linear, Boolean, Bayesian) will be discussed, as well as methods of inferring these from observed measurement data. A number of alternative network models more suited for high-level simulation of cellular behaviour will also be introduced. Building on the network inference methods, a number of ways of integrating various data sources and databases to refine biological networks will be discussed, with specific attention to the use of sequence information to refine transcription regulation networks. Finally, we will give some examples of algorithms exploiting the networks found to learn about biology, specifically for inspecting protein interaction networks and for finding active subnetworks.
In preparation for the course, please read the following primers on
Not all topics discussed in these primers will be used extensively in the
course, but if you find yourself severely lacking in a certain area it may
be wise to look up additional texts.
A folder containing handouts of the slides, papers to be discussed and a lab
course manual will be distributed at the start of the course. Electronic
copies of the course material can be found here.
After the course, you will write a short research proposal (e.g. for an MSc
student) on the application of one or more of the algorithms discussed
during the course to a problem you encounter in your own research. The
proposal should clearly state the background, the motivation, the problem
statement and a proposed solution. Most importantly, it should be realistic,
i.e. not require many years of work, large investments or magic to finish.
You can discuss your idea for the proposal with the lecturers during the
course. An example proposal (by Peter van Nes)
and a set of guidelines are available for download.
Please mail your finished proposal to Dick de Ridder (as a PDF). The deadline for submission is February 17, 2012. We will strictly adhere
to this deadline; if you require extension, you should contact us well
in advance. The proposal will be graded "fail" or "pass", with one possible
resubmission.
You can register for this course through the NBIC or
ASCI websites. The maximum
number of participants is 40, so register soon to be sure of a course seat!
Should the course be overbooked, PhD-students in the BioRange programme or
in the ASCI research school will be allowed access first.
The course is free for PhD-students working in or matched with the BioRange research
programme or BioAssist project, coordinated by the Netherlands Bioinformatics Centre
(NBIC); for PhD students in the Swiss Institute for Bioinformatics; and
for PhD-students in the Dutch ASCI, IPA, SIKS, OzsL, DISC or ImagO research
schools. For others, please refer to the NBIC/ASCI sites for fees.
The fee includes:
Information about hotel accommodation in Delft during this week can be
found here.
Participants have to book (and pay for) the accommodation
themselves if they need it. This is not included in the course fee.
One full week, followed by a final assignment.
Most days are laid out uniformly, roughly as follows:
Material
Examination
Registration
Format
| 9.30 - 12.30 | Lectures |
| 12.30 - 13.30 | Lunch |
| 13.30 - 15.30 | Participants read a scientific paper on the topic of the day, in small groups. Each paper is read by two groups, both of which prepare a short presentation (10-15 minutes). One of these groups is then randomly chosen to present, the other to lead the discussion. |
| 15.30 - 17.30 | A hands-on computer lab course on the algorithms discussed. |
| 1. Monday 16-1-2012 | Networks in biology |
| Lecturer | Dick de Ridder |
| Subjects | A brief overview of moleculary biology: DNA, RNA, proteins and metabolites. High-throughput measurement techniques and databases available. The role of networks in molecular biology. Examples of biological networks: regulatory programmes, signalling pathways and metabolic pathways. Network analyses: as graphs, as steady-state descriptions and as dynamical systems. Network properties (small world properties, hub; dynamic properties, stability). |
| 2. Tuesday 17-1-2012 | Network inference |
| Lecturer | Marcel Reinders |
| Subjects | Frequently used network models, focusing on transcriptional regulation: based on ODEs, Boolean logic or Bayesian statistics. Methods to derive network structures from data based on regularisation and subspaces. |
| 3. Wednesday 18-1-2012 | Network enhancement |
| Lecturer | Marcel Reinders & Dick de Ridder |
| Subjects | Methods to find frequently occurring subsequences (motifs) and summarise these in position weight matrices (PWMs), and the use of these motifs to refine transcriptional regulation networks. Ways of integrating other types of data to derive annotations and links, using distances, probabilities or kernels (focusing on protein interaction networks). |
| 4. Thursday 19-1-2012 | Network validation and execution |
| Lecturer | Nicola Bonzanni & Anton Feenstra |
| Subjects | A discrete approach to network modeling. Using Petri-nets as a formal network modeling tool, discrete and coarse-grained levels of cell constituents can be modeled in a discrete event fashion to understand network properties and behaviour at an abstract level. Applications to signalling and regulatory networks is discussed using 'real-life' examples. |
| 5. Friday 20-1-2012 | Network mining |
| Lecturer | Aalt-Jan van Dijk |
| Subjects | Protein interaction networks: interaction network evolution, reconstruction of ancestral networks, network alignment for cross-species comparisons. Interaction specificity, predicting protein interaction sites using network data and sequence data, correlated motif mining (interaction-driven vs motif-driven approaches). |
| Lecturer | Gunnar Klau |
| Subjects | Active subnetworks: integrative approaches for finding active, deregulated subnetworks. Basics in combinatorial optimization. Evaluation of three methods (jActiveModules, DEGAS, BioNet/heinz) on a large protein-protein interaction network and a realistic cancer microarray dataset. |