Introduction to BioinformaticsPractical Applications

1
Introduction to Bioinformatics/Practical
Applications

• Uma Chandran, PhD,MSIS,
• Department of Biomedical Informatics
• University of Pittsburgh
• chandranur_at_msx.upmc.edu
• 412-623-7841
• 10/20/08

2
My Background

• Bioinformatics Service
• UPCI
• Department of Biomedical Informatics
• Clinical Genomics Facility
• Runs expression, SNP and microRNA microarrays
• Bioinformatics tightly integrated with data
  analysis
• Expression, SNP, proteomic, integration of
  proteomic and genomic data

3
Outline of course

• What is Bioinformatics?
• Bioinformatics Concepts and Challenges
• Role in Pathology Cancer Biomarker Discovery
• Role in Translational research

4
What is Bioinformatics?

• http//en.wikipedia.org/wiki/Bioinformatics
• Application of information technology to
  molecular biology
• Databases
• Algorithms
• Statistical techniques

5
Bioinformatics in Pathology

• Cancer Biomarker discovery
• Prognostic markers
• Good v bad outcome
• Predictive markers
• Will patient benefit from a particular drug
  treatment
• Pharmacogenomic markers
• Effect of drug on tumor
• Genomic and proteomic technologies use
  bioinformatics approaches
• Potential to impact diagnostics

Nature 452548
6
Why should pathologists develop bioinformatics
skills?

• Pathology will see a gradual shift from cell
  morphology to molecular structure and function of
  cells as an adjunct to the diagnostic process.
  In the future, pathologists must be able to
  render a tissue diagnosis - most particularly for
  but not necessarily confined to tumorsthat takes
  advantage of not only morphologic observations
  but also related to genomic and proteomic testing
  Bruce Friedman, U Mich

7
What do these studies have in common?

• Use high throughput platforms to characterize
• DNA level single nucleotide polymorphisms
  (SNP), CN and LOH SNP chips
• Changes in mRNA profile -chips
• Changes in protein levels mass spec
• Tissue microarrays Staining many samples
  simulatenously
• microRNA microRNA chips

8
High throughput technology

• Ability to interrogate 1000s of genes/proteins
• Genomics
• DNA
• SNPs CN and LOH with disease
• RNA
• RNA profile with disease and normal
• Proteomics
• Mass Spec
• Protein Arrays
• Citations
• 2000 - 400 articles 2008 - 20000 articles!!!

9
Expression chips

• Chip
• Probes for genes
• Labeled samples
• Hybridized to chips
• Detection
• Quantitation
• Comparison

10
Bioinformatics (Biology Computation)

• Analysis
• Algorithms, statistics

• Advanced Analysis and visualization
• Annotation, pathways

• Databases to store and share, annotate
  information

Lincoln Stein just another tool, like the
microscope to study biology
11
History of Bioinformatics

• Computational Biology
• Margaret Dayhoff was pioneer
• 1965 - Created searchable protein databases
• 1982 DNA database GenBank
• Bioinformatics apps
• algorithms for assessing similarity
• What changes in proteins are tolerated?
• Search protein or DNA sequence for similarity

12
Bioinformatics at your desktop

• Studies conducted by clinicians/scientist
• Translational benchside to bedside
• Need interdisciplinary team
• Clinicians, bioinformaticians, statisticians
• Researchers needs a basic understanding of the
  methods

13
Biomarker studies

• Prostate Cancer
• Gene expression profiles in tumors, adjacent
  normals, donor normals, metastatic samples
• microRNA regulation
• Tissue microarray study to look at fatty acid
  synthase in prostate cancer
• Endometrial Cancer
• Expression profile of Early Stage Cancer versus
  Serous tumors
• Proteomic profiles
• Renal Cell Carcinoma
• Copy Number and LOH
• Glioblastoma
• CN and LOH studies
• Thyroid Cancer
• CN and LOH studies
• Lung Cancer
• Serum and tisue proteomic profiles
• Ovarian cancer
• Biomarker profile

Acharya et al JAMA (2008) 1574
14
Bionformatics Concepts 1

• Experimental Design

15
Experimental Design
SAMPLES

• Experimental Design is Critical!!!
• Chips are expensive
• What is the question
• Are there enough samples?
• Grade/Stage, Mets, Normal
• Row (genes), Columns (Samples)
• Consult with statistician before

G E N E S
16
University of Pittsburgh Molecular
Reclassification of Prostate Cancer Study

• 60 tumors, AN, Donor, 25 Mets
• ObjectiveIdentify expression profiles for tumor,
  normal, grade, stage, outcome, mets Results
• Tumors different from donor normals
• Field effect
• Mets also a different profile specific pathways
• Unequal distribution of grade stage
• Unequal distribution by outcome
• 4 Met patients but multiple samples
• Are Met patients different from tumor?
• 4 v 60
• Are metastatic samples different from tumor
• 25 v 60
• Profile for organ of metastatis?
• Not enough samples
• Comparison to other prostate studies
• Different normal used small studies
• Different distribution of stages/grades

17
Bioinformatics Concept 2

• Data Analysis

18
Need for computational methods

• Data Management
• Each file for a chip experiment is large
• 100MG x 10 1G
• Generates Gigabytes of data
• Data analysis
• 1000s of genes (or SNPs) and few samples
• How to find differences between samples
• What statistical methods to use?
• Like finding needle in a haystack

19
Analysis

• High dimensional data
• Data in proprietary formats
• How to open files
• How to analyze
• Are there commercial or academic software?
• Time-intensive, need dedicated time
• Statistical methods
• T tests, non parametric, others
• How to interpret results
• Fold change, p value, gene lists
• Pathways
• Resources
• Consult biostatisticians
• Bioinformaticians Core Service at UPCI
• HSLS offers software licenses
• Courses
• Microarray
• SNPs
• Statistics

20
Data analysis

• Class discovery
• Are there novel subclasses within data?
• Class comparison
• How are tumor and normal different in expression?
• Which SNPs are different?
• Class prediction
• Predict class of new sample
• Advanced pathway Analysis

21
Unsupervised analysis Class Discovery

• Are there novel subgroups that can be discovered
  based on expression profiles
• Need both analysis and visualization tools
• Hierarchical clustering, SOM, K means
• Principal component
• Challenge
• Discovery methods are borrowed from other domains
• Do not necessarily represent biological data

22
Data analysis Class comparison Expression (mRNA)

• Expression study
• Tumor v Normal Mets v organ confined
• Genes that are differentially regulated
• Statistics
• What are the underlying assumptions?
• Is it normally distributed
• How to set cutoff?
• Multiple testing correction

23
Single nucleotide polymorphisms (SNPs)

• Millions of variants
• Variants may be associated with disease
• SNP study
• Characterize in cancer
• Amplification, deletion (LOH), copy neutral
  deletion, Amplification and simultaneuous
  deletion
• Higher resolution than CGH and detect LOH and
  amplification on same data set
• Consult bioinformatics group

24
SNP chips to detect LOH

• Call are
• AB (heterozygous)
• BB/AA (homozygous)
• Compare tumor and normal samples
• Normal AB
• Tumor BB (or AA)
• This is LOH
• Is it in a known gene?
• Function of genes

25
SNPs to detect Copy Number changes
amplification
amplification
diploid
deletion
26
Challenges in SNP analysis

• Not many available tools
• SNP 6.0 measure about a million SNPs
• How to separate noise from real data
• Normal contamination
• How to confidently detect a region of copy number
  changes
• Statistical methods HMM, CBS
• How to link outputs to genomic information

27
SNP study

• Whole genome SNP arrays as a potential diagnostic
  tool for the detection of characteristic
  chromosomal aberrations in renal epithelial
  tumors Hagenkord et al
• Methods very different from expression arrays
• Again, experimental design key!!!!
• Which normal, paired or unpaired
• Sample size enough cases to answer question

28
(No Transcript)
29
(No Transcript)
30
Pathway Analysis
31
Bioinformatics Concepts 3

• Data Quality (Variation on Experimental design)

32
Molecular Markers - Expression

• Prognostic - breast cancer gene expression
  signatures
• Multigene recurrence expression signature
• ER with luminal A and luminal B subtypes
• ER with Her2 and basal subtypes
• Potentially predictive signature of genes
• Oncotype DX (Genomic Health), MammaPrint
  (Agendia) and the H/I test (AviaraDx)
• ER, lymph node negative patients
• Calculates a recurrence score
• 4000
• decide who should receive systemic therapy to
  eliminate any remaining tumour cells (that is,
  adjuvant therapy) after surgery, to reduce the
  risk of relapse.
• AMACR in prostate cancer
• SNP associated with gastric cancer PSCA gene

33
Why more biomarkers not in clinical use

• QC
• Tissue
• Experimental
• Clinical
• Small Study sizes
• Experimental design
• Analytical Methods

34
Variability in Tissues

• Sample QC
• tissue acquistion details
• Warm ischemia time,
• Storage medium
• Bulk, microdissection
• Sampling details, how far was tumor from normal
• What percentage of the sample was epithelial or
  stromal
• Experimental details such as quality of RNA,
  methods
• SNP studies
• FFPE v Frozen difficult to compare
• Samples from different lab are difficult to
  compare
• How to generate normals? Paired normals, lab
  normals, Hap Map normals

35
Experimental variability

• Expression
• Two rounds of amplification v single round
• Time for amplification
• Kits used
• Platform specific difference
• SNP
• Processing differences for frozen and FFPE
• fragmentation

36
Clinical information

• Correlation between profiling subgroups and
  outcome? Biomarker discovery
• Patient annotation
• Clinical variables not available not often in
  publications
• Even if author contacted, may not be possible to
  obtain
• Outcome -cancer registries
• Case information - EMR

37
Analytic methods many studies, many methods
Dupuy and Simon, JNCI 2007
38
Bioinformatics Concepts 4

• Infrastructure for translational research

39
Databases - Rich, highly annotated data sets

• Translational research benchside to beside
• deidentified clinical annotation, outcomes
• Tissue inventory and annotation
• Experimental data management
• Analysis tools

40
Challenges

• Institutional differences in tissue banking and
  access to deidetified Clinical info, Outcome from
  Registry
• Honest Broker, IRB
• Tissue annotation
• Different requirement for data elements
• Annotation of archival versus fresh
• Annotation for clinical trial
• Customizable
• LIMS for experimental
• Unlike traditional LIMS, dynamic every changing
  landscape
• Large data sizes, management challenges
• How to provide data to researchers, archival
• Need infrastructure and
• Analysis environment
• Integrate clinical and research data in a
  dynamic, customizable analysis interface

41
Translational research workflow
42
INTEGRATION VISION
Pathology Department
Cancer Registry
Tissue Bank
Clinical Gene Expression Labs
Proteomics Labs
CoPath LIS
IMPATH
caTissue
De-identification
Organ Specific Databases (of clinical and
outcomes data) Oracle 9i
Repository Annotated, deidentified clinical,
tissue, experimental
Analysis Interface
43
Biorepository- use this towards the end

• Pathology also must understand its stewardship of
  a resource that is likely to become more critical
  to bioinformatics research the tissue or
  biorepository. Says Dr. Becich
• Every tissue form surgical benches and every
  blood sample that comes into our clinical
  pathology laboratories has to be highly managed
  and refined to allow controlled genomic and
  therapeutic investigation
• Clinical annotation, processing annotations are
  critical

44
Clinical Annotation

• Deidentified integrated view of information from
  cancer registry and patient/tissue annotation
• Challenge
• Develop organ specific databases
• Multiple data source
• What do researchers and clinicians need
• is data available
• Develop standards

45
Data Sharing

• Need to get data to other investigators
• Consortia
• PI group
• Public
• Data are very large so need a way to exchange
  experimental data
• Annotate using standard vocabulary
• Clinical, tissue, experimental
• Provide a minimum set of annotations so that the
  data is useful and can be analyzed
• Interface for mining, analysis

46
National efforts to build a translational
research environment

• Need for the informatics infrastructure to
  facilitate translational research
• data management, analysis, data annotation,
  integration, exchange
• caBIG Cancer Bioinformatics Grid
• Enable translational research, biomarker
  discovery
• Grid computing, interoperable tools
• Many pieces have to come together in a workflow
• caTISSUE, caTIES, caArray standards based
• Clinical Translational Sciences Award (CTSA) to
  Pitt
• Pilot grants for translational studies where a
  researcher has to work with a clinician
• Example of studies expression and or/SNP with
  many diseases

47
Challenges

• Cancer Informatics Services
• Tissue Banking Tools
• Registry, Honest Broker
• Clinical Trials Management
• IT support
• Bioinformatics Service
• Clinical Genomics
• Clinical Protoemics

48
Conclusion

• Translational research and Bioinformatics are
  evolving rapidly
• Stay tuned!!