Bioinformatics in Cancer Biotechnology

1
Bioinformatics in Cancer Biotechnology

• Bob StephensAdvanced Biomedical Computing
  CenterAdvanced Technology ProgramSAIC-Frederick,
  Inc.National Cancer Institute at Frederick
• April 19, 2007

2
Objectives

• Overview/introduce bioinformatics concepts,
  applications and databases.
• Describe interplay between bioinformatics,
  technologies and the web.
• Profile importance of bioinformatics in cancer
  research.

3
What is bioinformatics ?

• Bioinformatics is the application of
  computational methods to the analysis of any type
  of biological data.
• Bioinformatics has become a diverse and
  multi-disciplined field that originally derived
  from computer science and biological science.

4
Evolution of bioinformatics

• Rapid technological advances in sequence
  determination set the pace for data acquisition.
• Similar advances in computing power and
  algorithmic approaches for sequence analysis,
  robotics enabled instruments.
• Co-evolution with web browser and programming
  language technologies.

5
Bioinformatics evolution (contd.)

• Additional high throughput technologies becoming
  available almost daily - microarrays, proteomics,
  population and genetic data, medical literature
  etc.
• Data volume is increasing at the same time as
  data complexity.
• Data distribution/synchronization becoming an
  increasingly difficult task.

6
Interplay between technology and bioinformatics

• New HT Technologies, eg. mRNA microarray
• Analysis and storage software
• Computational infrastructure
• Data integration

7
Example

• mRNA expression chip (20000 genes x 16 probes per
  gene), a few mb per sample.
• Data normalization software.
• Exon array - multiple probes for each exon for
  each of the 20000 genes - one file about 1gb.
• New normalization method requires all samples to
  be loaded simultaneously.
• More complex analysis reveals alternative
  splicing etc.

8
Interface of technologies and biology

• Experimental design very important in HT biology
• Experiments shaped by data access and
  availability
• Re-analysis of old data with new methods important

9
(No Transcript)
10
Bioinformatics historical perspective

• Stage 1 - bioinformatics term is coined to
  represent what had been DNA and protein sequence
  analysis (ca. 1995)
• Stage 2 - additional disciplines become rolled
  into bioinformatics including literature mining,
  statistical analysis, and virtually anything to
  do with computational analysis of biological
  data. (ca. 2000)

11
Bioinformatics - historical perspective (contd)

• Realization that bioinformatics is too broad a
  term, other disciplines break away eg. OMICs
  fields (eg genomics, proteomics others (ca.
  2001).
• Still later (current) realization is made that we
  wont be able to make any sense of individual
  disciplines without integrating them together,
  term now changed to integrative biology or
  systems biology (ca. 2003).

12
Importance of bioinformatics

• Bioinformatics has become a major part of both
  the NCI 2015 directive and the NIH Roadmaps.
• Virtually impossible to perform biological
  research without some form of computer aided
  analysis, especially in areas like genomics and
  proteomics.
• Important to keep scientific community in touch
  with developing technologies and capabilities for
  highest return on research investment.

13
Bioinformatics infrastructures

• Command-line implementations.
• Primitive GUI implementations.
• Sophisticated GUI interfaces and application
  packaging.
• Web interface and Java language gives platform
  independent access.
• PC-based, web-based and server-based
  architectures.
• Multiple tier infrastructures distributes
  computational burden.

14
What does bioinformatics technology involve ?

• Computer readable form of some type or types of
  biological data (instruments)
• Automation also requires programmable robotics
  capabilities (process science).
• Computer infrastructure for storing and analyzing
  the data.
• As data volume and complexity grows, the
  dependency on computer analysis increases.

15
Sources of bioinformatics technology

• Computer science leveraged technologies including
  algorithms and data representation models,
  visualization frameworks and programming
  languages.
• Web industry leveraged technologies including
  communication protocols, web servers and secure
  access.
• Database industry derived connectivity and
  technologies.
• Robotics and process engineering technologies for
  faster, cheaper throughput.

16
What can bioinformatics technology do for
biological science ?

• Develop uniform data standards and controlled
  vocabularies to allow for integration of
  disparate sources/types of data.
• Connect scientists to entire wealth of knowledge
  from basic science results to clinical trial data
  in context-sensitive manner.
• Fully integrate worldwide volume of knowledge,
  for example patient information
  disease-gttreatment-gtoutcome across multiple
  centers to allow for cross-comparisons.

17
NCI Resources

• caBIG NCICB Initiatives to develop integrated
  data/tool environment..
• Long term project requiring unprecedented
  cooperation, sharing.
• Short term solutions for day-to-day problems.
• Solution - use multiple approaches, staged
  implementation and layered technologies

18
(No Transcript)
19
ABCC hardware

• 128 cpu linux cluster (3.0 ghz processors).
• 256 cpu linux smp box with 1Tb memory.
• 64 cpu IRIX smp box with 256gb memory.
• 32 cpu IBM AIX smp computers.
• 16 cpu IBM HPC AIX smp computer.
• 8 x 8cpu IRIX computers.
• Other miscellaneous computers, disk storage, tape
  backup and network connectivity.
• Graphics visualization wall

20
(No Transcript)
21
ABCC Organization

• Networking and Security
• System administration
• Scientific program development
• Bioinformatics support
• Staff 40

22
ABCC Training Programs

• Classes for NIH/NCI scientists
• Unix, GCG, Java, High throughput sequence
  analysis, Geospiza (LIMS)
• Eudora, Advanced Eudora, Webmail
• Homology, Docking, QSAR, Intro to Modeling,
  Phred, Phrap, Consed
• One-on-one consulting services and training.
• Organize and host vendor specific training in
  genomics, pathways, and modeling

23
ABCC Support within ATP
Proteomics and Analytical Technologies (LPAT)
Computational Support Database Tools/Pathways Mass
Storage and Archive Pattern Analysis and
Clustering
Molecular Technologies (LMT)
Image Analysis (IAL)
Computational Support Database Tools and
LIMS Mass Storage and Archive Bioinformatics/Web P
attern/SNP Analysis
ABCC
Algorithm and Software Image Database Mass
Storage and Archive Viz Technology Development
Gene Expression (GEL)
Protein Chemistry (PCL)
Software Support
Gene Assembly and Validation
Animal Sciences (LASP)
Protein Expression (PEL)
Mass Storage Database
POET/Web
24
ABCC applications

• Sequence analysis - protein and nucleic acid, GCG
  and EMBOSS.
• Sequence assembly, SNP detection.
• Gene finders, analysis tools.
• Molecular modeling, docking.
• Molecular evolution and phylogeny.
• Computational chemistry.
• Linkage analysis.
• Proteomics.
• Classification tools (microarray and proteomics).

25
ABCC databases

• Genbank and derived divisions.
• Refseq, WGS, unigene divisions.
• dbSNP, gene, OMIM, homologene.
• UCSC, EBI and ncbi genome datasets.
• LIMS systems, data management.
• Uniprot, PDB, PIR, iProClass, Swissprot.
• CGAP, MGC data files, pathways.
• Medline, transfac and repeats data files.

26
ABCC web resources

• ABCC General information web page
  http//www.abcc.ncifcrf.gov
• ABCC account application information
  http//www.abcc.ncifcrf.gov/apps_apply.shtml
• ABCC Training web page http//www.abcc.ncifcrf.gov
  /training/courses.shtml
• ABCC scientific applications webpage
  http//www.abcc.ncifcrf.gov/app/htdocs/appdb/index
  .php
• ABCC GRID Database web page http//grid.abcc/ncifc
  rf.gov
• ABCC Pipelines web page http//www.abcc.ncifcrf.go
  v/app/login/login.php

27
The role of bioinformatics in cancer research

• Diagnosis - identify classifiers to better
  sub-divide cancer etiologies into groups. Better
  individual data to put treatment and individual
  together.
• Treatment - identify better methods to track
  treatment progress and indicate problems earlier.
• Prevention - understand mechanisms for cancer
  initiation, progression and development and
  identify targets in this process.
• Connect cancer patient data from geographically
  distributed cancer patients for more complete
  analysis.

28
Protein analysis tools

• Protein composition, isoelectric point, molecular
  weight analysis tools.
• Comparable alignment/searching tools for
  proteins.
• Protein secondary structure prediction tools.
• Protein structure modeling tools.

29
Genomics tools

• Gene finder and general genome annotation tools.
• Cross genome comparison tools and databases.
• Large scale sequence assembly and polymorphism
  identification tools.
• Genomic visualization tools (UCSC, NCBI,
  Ensembl).
• Data cleansing tools - vector screening, repeat
  masking.

30
Gene expression tools

• EST Clustering and differential expression
  analysis tools and databases.
• SAGE Analysis tools and databases.
• Microarray data collection, calibration and
  analysis tools and databases.
• Gene clustering and visualization tools.
• Integration tools - pathways, regulatory networks
  and medical literature.
• Databases for housing and querying the data.

31
Proteomics tools

• Mass spectroscopy tools for peptide
  identification.
• Fragment classification tools for identification
  of diagnostics
• Peptide fragment resolution tools -
  identification of protein mixtures from peptide
  sets.
• Databases for storing and querying the data.

32
Inherent bioinformatics problems

• Keeping data sources synchronized and up to date.
• Keeping applications up to date.
• Remaining aware of current palette of available
  tools and resources.
• Separation between computer developers and
  biologist users of software and databases.
• The silo concept- separate dysfunctional units.
• Lack of common language or database schema.

33
Data Analysis

• Pathway analysis
• Polymorphism
• Proteomics
• Image analysis
• Homology Modeling
• Live polymorphism analysis (if time permits)

34
Pathway Analysis

• Identify specific requirements of individual
  tumor.
• Advance to detection from diagnosis.
• Multiple points to cause aberrations and multiple
  points to act to correct them.
• Identify/characterize tissue, cell specific
  targets.

35
Pathway Gene Set Analysis

• Many experiments result in sets of genes, eg
  microarray, proteomics, literature searches etc.
• Clustering genes based on expression etc.
  provides only first dimension.
• View prospective pathways impacted by changes in
  expression, protein levels, phosphorylation etc.

36
G5G8Tg1Liver
G5G8Tg2Liver
G5G8-/-1Liver
G5G8-/-2Liver
G5G8-/-3Liver
37
G5G8Tg1Liver
G5G8Tg2Liver
G5G8-/-1Liver
G5G8-/-2Liver
G5G8-/-3Liver
38
Integrative Strategy for Microarray Analysis
Microarray Data
Clustering Analysis
Load into WPS
WSCP
Unassigned Genes
Integrate with WPS
Lists of Genes
Assign to uncharacterized pathway(s)
Assign to known pathway(s)
Putative Pathway
PSCP
PSCP
PSCP
39
Project Goal Integrate Biological Data and/or
Information Databases into Biological Networks
User input Microarray Data, Proteomics
Protein Interaction Database (BIND, DIP etc.)
Comparative Genomics
P1
P2
Protein Modification Phos., Glyco.
Gene regulation (Promoter etc)
Gene Ontology
SNP Haplotype Database (SNPinfo etc)
Literature DB (e.g. Pubgene ResNet)
NCBI resources OMIM etc

Statistical Evaluation Network Expansion (high,
low confidence)
40
One example of analysis scenario
microarray data pathway analysis or clustering
in local PC
Candidate gene sets Candidate pathway sets
Pre-computed DBs or Run-time computed
Internet-enabled
SNP Haplotype data (SNPinfo Disease association
Promoter Comparsion 1.CGI generator 2.CoreSearch 3
ConsInspector)
Protein interaction
Literature-based (Pubgene etc NCBI OMIM etc)
GO
Known gene training
Weighted scoring (Statistic analysis, filtering)
Final set of candidate genes (visualization and
re-creation of the new subnetwork within the
whole network)
Pathway expansion
41
Polymorphism Impacts

• Variation within species as great as differences
  between closely related species
• Confounds correlation analysis
• Impacts gene structure and expression
• Start with complete sequence for individual,
  obtain polymorphism data for populations/strains
  and breeds etc.
• Strains/breeds allow for good start

42
Polymorphism Types

• SNPs
• Indels
• STRs
• Tandem
• NonTandem (Copy number variation)
• Retroelement
• Complex
• Inversion/translocation

43
STR Polymorphism View
44
Strain Trace and Contig Coverage View
45
InDel Polymorphism Information View
46
Location Polymorphism Locator Query
47
STR Query results
48
Polymorphism Visualization
49
Proteomics InitiativeABCC Projects

• Disk Storage and Archiving (centralized storage)
• LAN Support
• Software Development
• Spectral Filtering
• Clustering/Biomarker Identification
• Database Development and Update
• Peptide identification DB
• MS Integration with Pathways
• ABCC Pathway tool
• Provide Scalable Computational Resources
• Software Optimization
• Sequest (working with LPAT,Yates Lab, and
  Thermoelectron)

50
Raw Data
Binning
Biological Marker
Clustering
51
Need for effective classification schemes for
correlating large amounts of data with Cancer
markers

• Large amounts of data.
• Many features (data points) to fit but few
  samples
• Problems are over-determined
• Solutions may be purely mathematical with no
  biological basis

52
Image Processing

• Confocal Microscopy Whole Animal Imaging
• 3D Segmentation
• Traditional/Real-time Microscopy
• Automated Quantitative Feature Analysis

53
Confocal Imaging

• Confocal Microscopy captures 3D volumes of tissue
  in situ
• Cancer appearance / development is related to the
  cellular neighborhood
• Therefore, segmentation and interpretation of
  cellular clusters is required
• NCI Developed Algorithms
• Segmentation needs human review

54
Imaging/Confocal Microscopy
55
Homology Modeling

• Many new chemotheraputic molecules are specific
  enzyme inhibitors
• Structural biology plays key role in
  design/enhancement of these compounds.
• Identify better inhibitors, understand specific
  differences and mechanisms.

56
Homology Modeling of Cysteing Finger in 3 Human
Raf Proteins
57
ABCC Bioinformatics Support Group

• Anney Che
• Jack Chen
• Jin Chen
• Qingrong Chen
• David Liu
• Uma Mudunuri
• Jigui Shan

• Wei Shao
• Gary Smythers
• Hong Mei Sun
• Natalia Volfovsky
• Xinyu Wen
• Ming Yi
• Jack Zhu

58
Bob Stephensbobs_at_ncifcrf.govwww.abc.ncifcrf.gov
Query tool
GBrowse