1. RAD (RNA Abundance Database)
Stoeckert, C.J.Jr., Pizarro, A., Manduchi, E., Gibson, M., Brunk, B., Crabtree, J., Schug, J., Shen-Orr, S., Overton, G.C.
A relational schema for array and non-array based gene expression data. Bioinformatics. In press. 2001
The Computational Biology and Informatics Laboratory

2. Issues Accurate experiment description Data preprocessing issues
clean-up
calibration
normalization
other transformations
Selecting, interpreting, and comparing experiments appropriately requires knowledge of how the experiments were performed and the samples that were used in sufficient detail to assess their quality and their degree of similarity.
The most appropriate criteria for spot selection, normalization, etc., depend on the experiments under study and on the questions investigated.

4. RAD Multiple labs Multiple biological systems Multiple platforms
Multiple image quantification software
RAD
Expressed genes
Differentially-expressed genes
Class discovery
Class prediction
Gene networks

5. RAD versatility Platforms 2-channel microarrays Filter arrays
Affymetrix
SAGE
Image quantification software
ScanAlyze
GEMTools
BioImage …

6. Views
A “view” renames attributes of a low-level generic table for specific implementations.
Common fields are specified as the same attributes for all implementations and implementation-specific fields rename generic attributes of the appropriate data type.
These views are not the same as materialized views that provide precalculated values to improve database query performance.

7. A SpotResult View

8. RAD strengths sample description use of ontologies
information about array elements
links to GUS (
storing of raw and processed data
captures all available information; history and parameter tracking
storing of public and proprietary data
user-group-other
read/write permissions
The schema is compliant with the minimum annotations recommended by MGED.
Controlled vocabularies:
Taxonomy: the Taxon table uses the NCBI taxonomy obtained from GSDB in relational form.
Anatomy: the Anatomy table is modified from the stage 28 (adult) mouse anatomy from the Mouse Gene Expression Database (MGD) at Jackson Laboratory to include human terms and expand description of systems such as hematopoiesis using medical textbooks as sources (e.g. the 37th edition of Gray’s Anatomy) as well as the expertise of the biologists at CBIL and in collaborating laboratories. The Anatomy hierarchy includes “cell lines” as a substructure for many different types of cells to distinguish immortalized from primary cells .
Disease: the Disease table uses the KEGG representation of the CDC ICD-9 classification. The KEGG representation has associated MIM identifiers with many of the ICD-9 terms.

9. Information to be captured
Figure from:
David J. Duggan et al. (1999) Expression Profiling using cDNA microarrays. Nature Genetics 21: 10-14

10. Categories of tables  Experiment Raw Data Platform Algorithm Metadata
Processed
Data

11. Experiment Tables B A Figure from:
David J. Duggan et al. (1999) Expression Profiling using cDNA microarrays. Nature Genetics 21: 10-14

12. Experiment Tables (A) Label Sample Treatment Disease Devel. Stage
Hybridization
Conditions
Label
Sample
Treatment
Disease
Devel. Stage
ExperimentSample
Taxon
Anatomy
RelExperiments
Exp.ControlGenes
ControlGenes
Experiment
ExpGroups
Groups

13. Experiment Tables (B) Views Experiment ExpImageImp ExpResultImp
PhosphorImager, ScanAlyzeImage, GEMImage,
StanfordScanner, AffymetrixScanner, SAGESequence, …
ExpImageImp
ExpResultImp
BioImage, ScanAlyzeAnalysis, GEMResult,
StanfordAnalysis, AffymetrixAnalysis, SAGEAnalysis, …

14. Platform Tables Figure from:
David J. Duggan et al. (1999) Expression Profiling using cDNA microarrays. Nature Genetics 21: 10-14

15. Platform Tables
SpotFamilyImp
SpotImp
Array

16. SpotFamily views (comparisons)
SAGESpotFamily
spot_family_id
tag
ext_db_id
cluster_id …
GEMSpotFamily
spot_family_id
ext_db_id
source_id
plate_id
plate_row
plate_column …
AffymetrixSpotFamily
spot_family_id
ext_db_id
accession …
Each is a view of SpotFamily table
Link to data with spot_family_id
Integrate through gene index (
GUS
EST assemblies
mRNA

17. Raw Data Tables SpotImp SpotResultImp SpotFamilyImp ExpResultImp
SpotFamilyResult

18. Processed Data/Algorithm Tables
SpotResultImp
raw spot value
SpotFamilyResult
summary of raw values
Algorithm
type of program used
AlgImplementation
actual program used
AlgoInvocation
usage of the algorithm
AlgParamKeyType
parameter data type
AlgParamKey
parameter description
AlgParam
value used
SpotResAnalysis
processed spot result
SpotFamResAnalysis
processed spot fam res
AnalysisType
type of processing

19. Query RAD by Sample or by Experiment
Access by Experiment
groups
Sample info
ontologies
Image info

22. What genes are expressed in the top 20% of normal
B-lymphocytes and mapped to Chromosome 19?

23. The allgenes (GUS) index provides annotation of array elements in RAD
EST clustering
and assembly
Different representations of the same RNA are identified.
EST/mRNA annotations are combined.
Consensus sequence is annotated (e.g., gene function).

24. GUS: Genomics Unified Schema
Ontologies GO
Species
Tissue
Dev. Stage
Genes, gene models
STSs, repeats, etc
Cross-species analysis
Genomic
Sequence
RAD
RNA Abundance DB
Characterize transcripts
RH mapping
Library analysis
Cross-species analysis
DOTS
Transcribed
Sequence
Special
Features
Transcript
Expression
Arrays
SAGE
Conditions
Ownership
Protection
Algorithm
Evidence
Similarity
Versioning
under development
Domains
Function
Structure
Cross-species analysis
Protein
Sequence
Pathways
Networks
Representation
Reconstruction

25. Different Views of RAD
Focused annotation of specific organisms and biological systems:
organisms
biological systems
Endocrine
pancreas
Human
Mouse
CNS
RAD
RAD
Plasmodium
falciparum
Hematopoiesis
*not drawn to scale*

28. Continuing Work and Future Issues
Analysis perspective:
ontologies
data preprocessing
cross-platform comparisons
utilize other types of high-throughput data (e.g. protein expression)
DB perspective:
capture conclusions from analyses in a structured way
integrate other types of high-throughput data

29. RAD: www.cbil.upenn.edu/RAD2
Elisabetta Manduchi
Angel Pizarro
Shannon McWeeney
Allgenes:
Brian Brunk Ed Uberbacher, ORNL
Jonathan Crabtree Doug Hyatt. ORNL
Sharon Diskin
Joan Mazzarelli
Jonathan Schug
EPConDB:
Greg Grant Klaus Kaestner, Penn
Phillip Le Marie Scearce, Penn
Debbie Pinney Doug Melton, Harvard
Alan Permutt, Wash U
MGED: