1. BDPGx - A Big Data Platform for Graph-based Pharmacogenomics Data
Pavan Kumar A
Big Data Analytics Team
C-DAC KP, Bengaluru

2. Outline Pharmacogenomics Biological Data Repositories
Graph Databases (What and Why)
Big Data Platform for Pharmacogenomcis Databases
Neo4j & Pharmacogenomics Graph Database
MapReduce : BLAST
Web Application for querying and visualization

3. Pharmacogenomics Pharmacogenomics = Pharma + Gene + Omics
Drug therapy consists of three major processes
Pharmacokinetic process
Pharmacodynamic process
Therapeutic process

4. Pharmacogenomics Pharmacogenomics led us Personalized Medicine
What and Why Personalized Medicine?

5. Pharmacogenomics : ADR
ERRORS’s in Health Care
Some Facts of ADR
1.6–41.4 % of patients undergo therapy prone to ADR’s
$17–29 billion spent annually to preventable ADR
In US, ADRs responsible for ~100,000 deaths annually

6. Pharmacogenomics : ADR
Factors for ADR’s
Genetic Factors
Pharmacokinetics
Pharmacodynamics
SNPs (Single Nucleotide Polymorphism)
Environmental Factors
Tobacco, Alcohol, Pollution, Diet habits and so on
Physiological Factors
Age, Gender, Disease state, Pregnancy, Starvation, Microbial Composition and so on

7. Pharmacogenomics : Pharmacokinetics
What the body does to drug.
This is captured by actions like
Movement of drug into the body, through the body and out of the body, which is referred as ABSORBTION, BIOAVAILABILTY, DISTRIBUTION, METABOLISM and EXCRETION

8. Pharmacogenomics : Pharmacodynamics
What the drug does to body.
This is captured by actions like
Receptor Binding
Post-receptor effects
Chemical Interactions

9. Pharmacogenomics : SNPs
Single Nucleotide Polymorphisms
Most common way type of Genetic Variation among people

10. Pharmacogenomics : SNPs
Diseases caused SNPs
Autoimmune Diseases
Genetic
Diseases
Cancers
Neurodegenerative Disorders
Cardiovascular
Diseases
Neuro-psychological
Neuro-psychological
Digestive Disorders
Addiction Dependence
Female-Specific Diseases

11. Pharmacogenomics : SNPs

12. Pharmacogenomics : Microbial Composition
Microbes in our body makeup to 100 Trillion cells (10 fold the number of human cells)
Image source: Digestive System Connection.htm

13. Protein structural variations Microbial Composition
Pharmacogenomics
Pharmacogenomics
Finally…
Protein structural variations
SNPs
Metabolomics
Gene Expression
Environmental factors: Chemicals, Diet, Tobacco, Alcohol etc
Physiological factors:
Age, Gender, Disease state, Pregnancy, Circadian rhythm, Starvation
Microbial Composition

14. How We Study ?
Data related to different domains are stored as Open Data Repositories
Download the data
Data Format : XML, CSV or Excel
Query a database via web application

15. Biological Data Repositories
Following are some of Pharmacogenomics Databases
PharmGKB – Pharmacogenomics Knowledge Base
DrugBank - chemical, pharmaceutical and pharmacological data
IGVdb - Indian Genome Variation Database
CTD - Comparative Toxicogenomics Database
STITCH (Search Tool for Interactive Chemicals) – Chemical Protein Interaction Networks
TTD - Therapeutic Target Database
KEGG (Kyoto Encyclopaedia of Genes and Genomes)

16. Integration of Biological Data Repositories
Data is spread across many repositories.
User has to navigate many pages on the web or across many websites.
So there is a need to integrate all the data to get consolidated information on place

17. Interlinked Biological Data
Databases
Consortiums
Tools
Information from Articles, Literature
Pasha and Scaria etal Omics for personalized medicine

18. Integrating Databases
Integrating many databases based on Internationalized Resource Identifiers (IRI)
Sample for SCN5A(Sodium channel protein type 5 subunit alpha)
Database
Gene
Organism
Len
Interacting
Chemical
Disease/
Disorder
Pathway/s
Chrom_Start
Chrom_End
Uniprot
SCN5A
Human
2016
CTD
SCN5A
sodium arsenite
Atrial
Fibrillation
Developmental Biology
PharmGKB
SCN5A
Database
Gene
Organism
Len
Interacting Chemical
Disease/ Disorder
Pathway/s
Chrom_Start
Chrom_End
My_DB
SCN5A
Human
2016
sodium arsenite
Atrial
Fibrillation
Developmental Biology

19. NoSQL database family

20. Graph Databases Graph Databases are NoSQL databases Family.
Pictorial representation of data in the form of Nodes and Edges (with or without properties)
Image Source :

21. Why Graph Databases? Ref : https://neo4j.com/use-cases/
Graph Databases are well suited for interconnected data.
Some of the use cases of Graph Databases
Fraud Detection
Graph-Based Search
Network and IT Operations
Real-Time Recommendations Engines
Social Network
Identity and Access Managements
Ref :

22. Graph Databases : Properties
Two important properties of graph databases technologies
Native Graph Storage
Some serialize to RDMS
Native Graph Processing (a.k.a “index-free adjacency”)
Connected nodes physically “point” to each other

23. Graph Databases

24. Graph Database : Neo4j
Most of the Biological data is interconnected, Graph databases are well suited.
World’s Leading Graph Database :
Open Source and Welcoming UI
Native graph storage with Native GPE(Graph Processing Engine)
Easy to represent connected data
Faster to retrieve/traversal/navigation of more Connected data
Represents Semi-structured data

25. Graph Database : Neo4j
In Neo4j, Cypher Query Language (CQL) is used to create nodes, labels, edges and properties
Example:

26. Pharmacogenomics Graph Database

27. Pharmacogenomics Graph Database

28. Pharmacogenomics Graph Database

29. Pharmacogenomics Graph Database

30. Pharmacogenomics Graph Database

31. BDPGx - A Big Data Platform for Graph-based Pharmacogenomics Data
Tools and Technologies:

32. BDPGx - A Big Data Platform for Graph-based Pharmacogenomics Data
BDPGx has
Nodes
Properties
Relationships
15 Relationship types

33. Conclusion
Biological data is generated from various sources and available in different formats
Finding correlations among the available data can give better insights
BDPGx
User-friendly access to get most appropriate information to the researcher

34. THANK YOU