Course Overview
- Introduction
- Cluster Detection
- Cluster Evaluation
- Cluster Visualization and Interpretation
50/50 Lectures and Practicals
Goals: Introduction
- What is clustering? Why cluster? Role in data integration
- Motivating biological and non-biological examples
- Approaches to clustering
- Overview of lab data set and practicals
Clustering in a nutshell
Some way to group elements that are more similar to each other than they are to everything else
What does it mean to be more similar?
Many approaches to clustering
Many others
Different approaches. Different metrics.
Distinguishing features of clustering methods
Hard Clustering
Elements belong uniquely to one cluster
Soft (Fuzzy) Clustering
Elements belong to more than one cluster
Clustering Nodes vs Edges
Nodes: Spectral clustering
Edges: Link-community clustering
Why cluster a network?
Define topologically interesting substructure
Why cluster a network?
Clustering is a way to decipher network structure
Hairball ⇨ Comprehension
Clustering for Data Integration
Kernels derived from 5 sources of information
Similarities between 4567 human genes
Selected from their relationship to 120 genes involved in mitosis, DNA mismatch repair, and BMP signaling
Course goals
- Combine graph kernels
- Cluster integrated network
Cluster detection, analysis and visualization
An Introduction
Author: Aaron Brooks / @scalefreegan
You can follow along on
http://scalefreegan.github.io/Teaching/DataIntegration
