Analysis of high grade prostate cancer microarray data

Analysis of High Grade Prostate Cancer Microarray Data

BIN714 Final ProjectGungor Budak

June 4, 2015

Instructor: Assoc. Prof. Dr. Yesim AYDIN SON

Outline

❏ Introduction & background

❏ Data description

❏ Experimental design

❏ Methods

❏ Results

❏ Conclusion

2

Introduction & Background

❏ In males, Below bladder and in front of

rectum

❏ In males, Contains cells that produce semen

Image: www.roboticoncology.com 3

Introduction & Background

❏ T2 stage

❏ only in prostate

❏ large enough in DRE

❏ T4 stage

❏ fixed or growing into

nearby structures

❏ N1

❏ spread to lymph nodes

❏ M1

❏ distant metastasisImage: www.cancerrecovery.org.uk

4

Data Description

Goal

Identification of diagnostic markers and

targets for novel therapeutic drugs for high

grade prostate cancer (PC) (Shuin et al., 2010)

ID Organism Type Platform Sample # (Can.)

Sample # (Nor.)

GSE45016 Homo sapiens

Expression profiling by array

Affymetrix Human Genome U133 Plus 2.0 Array

10 1

5

Experimental Design

❏ 10 frozen specimens with high PSA1 levels

and high Gleason scores (8-9), staged T2

to T4 with or without N1 and M1

❏ Normal prostate (NP) epithelial cells from

five non-prostate cancer (BPH2) patients

(males & mixed)1 Prostate-specific antigen2 Benign prostatic hyperplasia

6

Methods

❏ Data analyzed with GEO2R tool

❏ Log transformation applied

❏ eBayes feature selection

❏ Results (diff. expressed genes) filtered

❏ p-value < 0.05

❏ LFC > 2

❏ Only characterized genes (No LOC123456789)

❏ 1166 genes collected

7

Methods

❏ PC related genes collected

❏ KEGG Diseases (12)

❏ The GeneCards Human Gene Database (20)

❏ Dong JT, 2006 (30)

❏ DAVID web service used for functional

annotation (Dennis Jr et al., 2013)

8

Methods

❏ PCSF network generated (Tuncbag et al., 2013)

❏ LFC as “prize”

❏ Cost per edge, penalty per fail to include a node

❏ iRefWeb ref. interactome used (Wodak et al., 2010)

❏ Down to 549 genes

9

Results: GO Bio. Proc.

10

Results: KEGG Pathways

11

Results: Network Analysis

PC Genes Steiner Nodes Steiner PC Nodes

5/47 98/549 PTEN, TP53, BRCA1, GSTP1, ELAC2

12

Results: PTEN

❏ Phosphatase and tensin homolog

❏ Many frameshift

deletions

❏ Metastasis

❏ Best studied in PC

Vishwanatha et al., 2012, J Carcinog13

Results: TP53

❏ Tumor protein p53

❏ Commonly single

point mutations

❏ Most frequently

mutated in human

cancerBrosh & Rotter, 2009, Nature Reviews Cancer

14

Conclusion

❏ Analyses revealed PC related genes

PTEN & TP53

❏ Improvements

❏ More complete interactome

❏ Better experimental design

15

Thank you