BCH3031 Lecture Notes - Lecture 17: Hierarchical Clustering, Dimensionality Reduction, Interactome
25 May 2018
School
Department
Course
Professor
Lecture 17 – Part 2
Concept Networks in Biology
• Networks are connected by Nodes and Edges
o Nodes are genes or proteins
o Edges are interactions (genetic, transcriptional, physical)
• Some people like to think of analogies with social media networks
o Reduce system to a size we can think about
• Types and functions of Networks
o Depicting interaction network in yeast, First hub: RNA Pol II and
second: mRNA 3’ end processing machinery
▪ Blobs are proteins (nodes)
▪ Lines represent interaction between proteins
• Interactions can be experimental or curated for
databases and others
o Network usually help reduce complexity
o Nodes are the proteins/genes
o Edges are the interactions
o Multiple international efforts to consolidate systems data into
interactive formats
• Clustering of microarray data across multiple experiments
o Hierarchical clustering of gene expression
o Microarray of 4803 genes
o 7186 different experiments
o Genes with similar function cluster together
▪ Measuring RNA content within cell
▪ Transcriptional wiring
▪ Cells go through metabolic cycle → bursts of transcription over
time
o Spell dataset
o Dimensionality reduction (t-SNE)
▪ Microarray of 4803 genes – to reduce complexity
▪ Transcripts with similar regulation cluster together
• Two-dimensional hierarchical clustering analyssi of Chlp-chip data
o Hierarchical clustering of transcription factor binding sites
o 288 experiments
o 155 DNA binding proteins
o Genes co-regulated by the same DNA bind protein cluster together
o Re-analysis of Harbison et al 2004
• Protein complex network of Dresophila melanogaster
o Protein associations involving nearly 5,000 individual
o 556 protein complexes
o Assign functional links to 586 unannotated genes
How Networks help us Understand Gene/Protein Functions
• Open source database mining
• Information gathering from available data to refine new hypotheses
• New biomarkers or therapeutic targets for disease
• Mining networks for new insight:
find more resources at oneclass.com
find more resources at oneclass.com
