Transcript Architectural

Phase II Additions to LSG
• Search capability to Gene Browser
– Though GUI in Gene Browser
• BLAST plugin that invokes remote EBI
BLAST service
• Working set manager
– State retention between sessions
• Provenance viewer
– Displays annotated provenance
– Accepts RDF representations
Knowledge Discovery through Provenance
Collection, Representation, and Use in the
Life Science Grid (LSG)
Phase II Final Report : Architectural and
Technical Details
Beth Plale
Director, Center for Data and Search Informatics
Indiana University
Key contribution to LSG
proper (provenance aside)
Introduction of state
LSG Space
user
Annotated Provenance Graph
Working
Set
Manager
RDF
Viewer
BLAST
OPM* RDF Interface
Provenance
DB
Karma Services
Gene
Browser
Lilly to Karma
Events Reflector
Gene
Ontology
Lilly CAB Bus
Karma Framework
Entrez
Provenance
Graphs
S-OGSA
Service
Semantic
Binding
Events Capture
Proxy
Karma
structure
ontology
Karma Events Bus (WS+LSG)
Services and data
ontology (myGrid)
SAWSDL
Registry
Resources
(public + private)
Service Annotations
*Open Provenance Model v1.01
BLAST support
Working Set: support for user state
Demo 1: Phase II Use Case
• Select “gene” from database list, list will
show in Gene Browser
• Submit gene to NCBI
• Open tab of BLAST plugin, download
FASTA sequence
• Run BLAST
• Add results to Working Set
• Annotate Working set
Working Set Manager listens to
CAB bus. It uses Entrez ID and/or
all or partial of BLAST result as
input to working set, or imports csv
file into working set. Working set
WS2 was generated from working
set WS1 by Delete Rows. WS2
can be exported as csv file.
Demo II: Query provenance database
Query BLAST related data.
Query 1: get the latest Blast_Plugin.
create or replace view v1 as
select process_id, service_id, process_initialization_time
from process
where service_id like '%Blast_Plugin' and
process_initialization_time = (select max(process_initialization_time)
from process
where service_id like '%Blast_Plugin’)
select * from v1;
Query 2: get the service (Blast_Ebi_Web_Service) invoked by Blast_Plugin.
select invoker_id, invokee_id, p.service_id
from invocation, process p, v1
where invoker_id = v1.process_id and invokee_id = p.process_id;
Demo II: Query provenance database
Query 3: get input to Blast Ebi Web Service
select p.service_id, artifact_id,artifact_value
from artifact_used au,artifact a, process p, v1
where au.artifact_no = a.artifact_no and
au.process_id = p.process_id and
p.process_id = v1.process_id + 1 and p.service_id like
'%Blast_Ebi_Web_Service’;
Query 4: get output from Blast Ebi Web Service
select p.service_id, artifact_id,artifact_value
from artifact_generated ag,artifact a, process p,v1
where ag.artifact_no = a.artifact_no and
ag.process_id = p.process_id and
p.process_id = v1.process_id + 2 and p.service_id like
'%Blast_Ebi_Web_Service’;
Suggested Future Work
• Next steps:
– Engagement of users or,
– Expanded functionality set
• Build in development time if we must do this ourselves
• Represent to user combined visualization and
process provenance
• Write research quality paper
– Requires user study, or comparison, or …
• Formally integrate provenance collection tools
into non-public LSG
Suggested Future Work:
Technical
• Support BLAST in asynchronous mode,
• extend NCBI Entrez to work on other NCBI
databases, and
• design rich provenance queries.