858L: Homework #1 Fall 2009
Due: September 21, 2009, by the start of class.
For this assignment, you should write two programs in the Python programming language. You
should work in groups of 2. If you are not familiar with Python, you can nd a tutorial h
858L Homework #2
Due: October 19, before class.
Modularity. Let G = (V, E) be a simple, undirected graph with adjacency matrix A. That is
Auv = 1 if edge cfw_u, v E and 0 otherwise. Let ku be the degree (number of neighbors) of a vertex
u. If we partition
with Saket Navlakha, James White,
Niranjan Nagarajan, Mihai Pop
Graph summarization creates a
hierarchical decomposition of
Many other network
decomposition methods are
based on hierar
In silico prediction of edges
Even high-throughput experiments are expensive
Look to computation to infer linkages
Experiments are noisy, so the bar is low :)
Combination of multiple experiments improves accuracy drastically;
stands to reason
CMSC 858L Lecture 6 7
September 23, 2009
Slides by: Saket Navlakha
High-throughput methods are
producing lots of protein
interaction data for many
Predicting Protein Function from Networks
Ultimately, we want to know how various processes
in the cell work.
A rst step: gure out which proteins are involved
in which biological role.
What do we mean by a biological role?
(and other clustering algorithms)
Comparing Clustering Algorithms
Brohee and van Helden (2006) compared 4 graph clustering
algorithms for the task of nding protein complexes:
RNSC Restricted Neighborhood Search Clustering
SPC Super Paramagn
Transcription Factor Networks
Transcription factors = class of proteins
that bind to DNA to regulate genes.
Nodes represent proteins, directed edges
TF often regulate other TF.
Hedgehog response pathw