What can semantics do for Bioinformatics?

1
What can semantics do for Bioinformatics?

• Semantic approaches to Bioinformatics

2
Outline

• Challenges in biology
• What can bioInformatics do for Biology?
• What can Semantics do for bioInformatics?
• The field so far
• Challenges

3
Challenges in biology

• What makes us ill or unwell?
• Disease identification, disease inducing agents
• What keeps us healthy and makes us live longer?
• Drug discovery
• Where do we all come from and what are we made
  of?
• Genetics

4
and their implications

• Understand biological structures of increasing
  complexity
• Genes (Genomics)
• Proteins (Proteomics)
• Carbohydrates (Glycomics)
• Understand biological processes and the roles
  structures play in them (biosynthesis and
  biological processes)

5
Outline

• Challenges in biology
• What can bioInformatics do for Biology?
• What can Semantics do for bioInformatics?
• The field so far
• Challenges

6
What can BioInformatics do?

• Analyze genetic and molecular sequences
• Look for patterns, similarities, matches
• Identify structures
• Store derived information
• Large databases of genetic information

7
(No Transcript)
8
Outline

• Challenges in biology
• What can bioInformatics do for Biology?
• What can Semantics do for bioInformatics?
• The field so far
• Challenges

9
What can Semantics do?

• Semi-automated annotation of resources of various
  sources.
• Creation of knowledge bases from the annotated
  resources.
• Relationship discovery among the annotated
  resources and entities
• Both implicit and explicit relationships
• Inter and intra ontology relationships
• Unified access to multiple sources

10
Unified Access to Multiple Sources

• Automated and semi-automated generation of web
  services
• Too many different sources with different access
  rules available

11
?
Heterogeneous data sources on the web
12

• Examples of studies involving carbohydrates
• - how does the glycosylation pattern of the cell
  change when it undergoes a physiological change
  such as differentiation or transformation to a
  cancer cell?

13
Applying semantic techniques in Drug Discovery
14
Behavior of disease causing agents
Pathogen Classification
Receptor Binding
Non Receptor Binding (Cell attackers)
Receptor Binding Use Enzymes to bind to
receptors (Human system elements to which the
disease causing agents bind) eg - HIV protease
Non receptor binding These attack the cells
directly Eg - Viruses
15
How chemists and drug developers work?

• Identify the chemical composition of the pathogen
  from the wet-lab experiments on the pathogen.

16
How chemists and drug developers work?

• Discover potential receptor sites or cells in the
  human system susceptible to this pathogen.

17
How chemists and drug developers work?

• Model the chemicals that can poison the
  pathogen at the discovered receptor site or cell.
  This is the DRUG !!!

18
The facts

• The proportion of industry RD expenditure
  allocated to clinical studies increased between
  1996 and 1998 from 32.5 to 39.5, respectively.
  1
• Time for a drug to reach the market 6 YEARS.
• Large of number of potential drugs tested do not
  show predicted behavior
• A few samples reach the stage of clinical testing
• The longer the drug stays in the lab, the
  expensive it becomes when it is marketed.

19
The Ultimate Objective

• Discover better drugs
• Faster and cheaper to the
• common man

20
Adding automation and intelligence Can realize
this objective
21

• Drug-Ontology
• Compounds
• Mol.composition
• Reaction types
• properties

• Recep-Ontology
• Receptors
• Chemical comp.
• Enzymatic state
• Cell
• Susceptible
• molecules

• Patho-Ontology
• Mole. Composition
• Suscep. receptors

Chemical
Ontology Lookup
Semantic Query Engine
Result Document
Query Compound Sildenafil Citrate Ontology Compoun
ds
22

• Drug-Ontology
• Compounds
• Mol.composition
• Reaction types
• properties

• Recep-Ontology
• Receptors
• Chemical comp.
• Enzymatic state
• Cell
• Susceptible
• molecules

• Patho-Ontology
• Mole. Composition
• Suscep. receptors

Ontology Lookup
Intra Ontology Relationship
More description About inter and Intra ontology
Relationships will Come on this slide
A sample ontology Using DAMLOIL will Be
captured in this slide Or in a new previous slide.
Inter Ontology relationships
Semantic Query Engine
Result Document
Query Compound Sildenafil Citrate Ontology Compoun
ds
23
The Result
headache
has_side_effect
3.5 mg/mL
4-Methylpiperazine
has_solubility
is_made_up_of
phosphodiesterase type 5 (PDE5)
Sidenafil Citrate
Pyrazolo pyrimidinone
is_made_up_of
action_inhibits
commercial_name
There will be another slide Describing the
relationships using DAML OIL
Viagra
24
Discovery by EliminationLet the agents to the
work
25
Discover by Elimination

• Some molecules contain compounds that inhibit
  them from acting as drugs
• Elimination of these groups can make the
  molecule, a DRUG.

26
Can Molecule X be used as a drug to cure Disease
D?
27
Step 1 Capture domains using ontologies.
DISEASE ONTOLOGY
Disease D

• Pathogen
• X

PATHOGEN ONTOLOGY

• Compound
• P

Pathogen X
Compound Q
28
Step 2 Traverse explicit relationships.
DISEASE ONTOLOGY
STEP 1 1.Look up the disease ontology 2.
Identify the disease causing pathogen.
Disease D
STEP 3 1.Look up the molecule ontology 2.
Identify the composition of the possible
drug.

• Pathogen
• X

STEP 2 1.Look up the pathogen ontology 2.
Identify the molecular composition of the
pathogen.
PATHOGEN ONTOLOGY

• Compound
• P

Pathogen X
Compound Q
29
Step 3 Discovering Implicit relationships
INHIBITS
Extract the relationships amongst the
compounds of the potential drug and the
pathogen.
PATHOGEN ONTOLOGY
INHIBITS

• Compound
• P

Pathogen X
Compound Q
PRODUCES TOXIN
CAUSES SIDE EFFECTS IN DISEASES CAUSED BY
PATHOGEN X
30

• C doesnt contribute to the curing aspect of the
  drug, but rather generates a toxin.
• Eliminate C and molecule A can be a potential
  drug.
• Eliminate and Discover!!!!

31
Get the right drug Not the side effects
32

• Side Effect An unfavorable response to a drug,
  caused by reaction of drug with an existing
  physical condition other than the disease.

33

• Drug-Ontology
• Compounds
• Mol.composition
• Reaction types
• properties

• Recep-Ontology
• Receptors
• Chemical comp.
• Enzymatic state
• Cell
• Susceptible
• molecules

• Patho-Ontology
• Mole. Composition
• Suscep. receptors

Ontology Lookup
Intra Ontology Relationship
Inter Ontology relationships
Semantic Query Engine
Result Document
Drug A, Disease X, Other diseases Molecule
Onto,Pathogen Onto, Receptor Onto
34
DISEASE ONTOLOGY
MOLECULE ONTOLOGY
Disease X
Molecule A

• Compound A

Compound C
Disease Y
Compound B
Reacts_unfavorably
Inhibits_molecule
PATHOGEN ONTOLOGY
Q
R
35

• Molecule A cures disease X, but at the same time
  reacts unfavorably when the system has another
  disease Y.
• This relationship can be discovered and person
  with disease Y will not be administered drug X.
• Just the DrugNo Side Effects !!!

36
Where can semantics help?

• A semantic data store of disease causing
  organisms and their properties and structures.
• Any organism can be then compared with this to
  get the type of disease they will possibly cause.

37
References

1. Kermani F and Findlay G (2000). The
   Pharmaceutical RD Compendium. http//www.cmr.org
   
2. Anon. AstraZeneca Unveils Promising Portfolio
   with 57 NCEs. Pharma Marketletter 20 December
   2000 p18.

38
Common Process

1. Use GO ID corresponding to the known gene as
   input to step 2.
2. Retrieving these similar sequences using Entrez
3. A multiple sequence alignment using CLUSTAL
4. Constructing phylogenetic tree using PAUP

39
Our Approach automation

• Decompose query into semantically annotated
  workflows using METEOR-S
• Semantic Templates at each node allow choosing
  multiple tools automatically
• Run multiple instances of workflows using
  different tools, combine results

GO id
SIMILAR SEQUENCES MATHCER
MULTIPLESEQUENCEALIGNMENT
PHYLOGEN TREECREATOR
40
Semantic Bioinformatics Processes
DATA SEMANTICS
Use Concept Ontologies for finding relevant
services
GO id
SIMILAR SEQUENCES MATHCER
MULTIPLESEQUENCEALIGNMENT
PHYLOGEN TREECREATOR
41
Semantic Bioinformatics Processes
FUNCTIONAL SEMANTICS
Use Functional Ontologies for finding relevant
services
GO id
SIMILAR SEQUENCES MATHCER
MULTIPLESEQUENCEALIGNMENT
PHYLOGEN TREECREATOR
42
Semantic Bioinformatics Processes
QoS SEMANTICS
Use QoS Ontologies for finding relevant services
GO id
SIMILAR SEQUENCES MATHCER
MULTIPLESEQUENCEALIGNMENT
PHYLOGEN TREECREATOR
43
Semantic Bioinformatics Processes
EXECUTION SEMANTICS
Use Execution Semantics for execution monitoring
GO id
SIMILAR SEQUENCES MATHCER
MULTIPLESEQUENCEALIGNMENT
PHYLOGEN TREECREATOR
44
Semantic Web Process LifeCycle
Description / Annotation
Execution / Orchestration
WSDL, WSEL DAML-S Meteor-S (WSDL Annotation)
BPWS4J, Commercial BPEL Execution Engines,
Intalio n3, HP eFlow
Semantics Required for Web Processes
UDDI WSIL, DAML-S METEOR-S (P2P model of
registries)
BPEL, BPML, WSCI, WSCL, DAML-S, METEOR-S (SCET)
Discovery
Composition / Choreography
45
Semantic Querying
46
GNT-V Upregulate in meta-static Cancer cells is
here
47
Proteins involved in N-glycan synthesis is here
48
Almost all (N) Glycoproteins will be Found here.
49
Lectins that bind various families of N-Glycans
will be found here. Possible interference to
data reg. Biological (disease) studies
where Lectins were used as a Diagnostic or
biochemical Probe.
50
(No Transcript)