Bioinformatics Pipelines for Analyzing ISRCE Datasets to Answer Key Systems Biology Questions课件.ppt

Bioinformatics Pipelines for Analyzing ISRCE Datasets to Answer Key Systems Biology Questions Yuriy Gusev Clinical Research Informatics Georgetown/Lombardi Bioinformatics Analysis Pipeline Raw Data Sets: public or local Gene Expression, DNA copy number, microRNA expression … Data Proccessing: Uniform Signal Extraction, Normalization, Quality Control Group Comparison Based on Clinical Attributes: Statistical Tests, Cluster Analysis, Principal Components Analysis, Classification (PUG-SVM) Advanced Analysis: Functional profiling; Pathway Analysis, Network Analysis, Machine Learning /Classification Biological Interpretation Experimental Or In Silico Validation Novel Hypothesis Study/Data Inventory 1 Loi et.al. 77 ; 741 2 Wang et.al. 286 3 Desmedt et.al. 55 4 Zhang et.al. 136 5 Zhou et.al. 54 6 Sotiriou et.al. 189 7 Edinburgh In progress Breast Cancer Studies Total Cases: 1538+ Study IDs Study_Name Loi et.al. ? Pubmed_ID ? GEO_ID GSE9195; GSE6532 ? ArrayExpress ID E-GEOD-9195; E-GEOD-6532 ? Platform_ID/Chip_Type H-GU133A ; H-GU133A, H-GU133B, H-GU133_ Plus_2.0 CLINICAL ATTRIBUTES Number of Samples 77 ; 741 Cancer Type Breast_Cancer ? Tissue Type Breast_Tumor ? PhenoData Age, ERS, LNS, PRS, TG ,TS ? Missing Attributes NA ? Treatment Type TAM PRE- PROCESSING Background Correction Standard non-linear correction Normalization RMA (Robust Multichip Analysis) Quantile Normalization ? Summarization Median polish ? QC/QA Subset H-GU133A Case Study: Loi et al. Additional Studies: Assessment in Progress (Rebecca Riggins) Case Study (Loi et al.): Differential Gene Expression Clinical Outcome: five year DMFS Data Analysis Details: Two options for processing: Original Affymetrix Probes Sets: Multiple Probe Sets per gene Custom Chip Definition Files (CDFs): Single Probe Set per gene (Bioconductor) Quality Control : Checking for outliers Group Comparison: DMFS5years vs DMFS5years T-test with multip