goalnet: intelligence in workflow orchestration zhiqi shen nanyang technological university june,...

Post on 21-Jan-2016

217 Views

Category:

Documents

0 Downloads

Preview:

Click to see full reader

TRANSCRIPT

Goalnet: Intelligence in Workflow Orchestration

Zhiqi Shen

Nanyang Technological University

June, 2007

Background

• Web services support interoperable machine to machine interaction over the internet;

• Workflows describe tasks and execution procedures;

• Business process modeling; and

• System integration.

Challenges

• Web services are not always available;

• Web service needs to be selected dynamically;

• Many workflows available defined using different tools;

• Process execution path is decided on the runtime; and

• Decision making is required.

Our Solution

• Goal oriented modeling;

• Decision making on goals and actions;

• User interaction;

• Agent mediation and monitoring;

• Goal Net modeling tool– Process modeling– Agent modeling

Overview of Goal Net

Root Goal

Atomic Goal

Transition

Composite Goal

Intelligence for Decision Making

• Process modeling

• Web service/workflow orchestration

• Goal selection and action selection

• Multiple decision making mechanisms support

• User interaction

Agent based System

• Agent system generation

• Development platform available– Goal Net Designer

– Agent Creator

– Goal Net Loader

Root Goal

Atomic Goal

Transition

Composite Goal

GoalNet Designer - Process Modeling Tool

Workflow/Web Service Orchestration

• Tasks in transitions– Workflows– Web services– Functions

• Services selected by agents during runtime

• Execution path (goal pursuit) is decided at the runtime

Proposed Architecture for The Bio-Manufacturing Domain

Grid/web services

Grid/web services

Bio-Manufacturing Grid

Objective is to reduce time from biomedical research to mass production of diagnostic kits.

The system integrates workflows and supply chain processes using the Grid

• Collection, storage and tracking of bio-samples

• Sequencing of the genomes of samples

• Design and Fabrication of DNA chips to test samples

• Analysis of results

• Making cost effective diagnostic kits via genes identified

New Requirement for Bio-Manufacturing

Individual Low-level Bio-Systems are available

Existing Tools for Low-level Workflow Orchestration - Taverna

Existing Tools for Low-level Workflow Orchestration - Kooplatform

Grid based Bio-Manufacturing

Key IT components:

1. GoalNet: Developed at MC3/NTU choreographs grid workflow engines and processes.

2. KOOPlatform: Developed by KOOPrime for orchestration of laboratory and bioinformatics processes

3. FPGA card, a system developed by Progeniq

4. Taverna, the leading European bio-grid workflow system developed by MyGrid project

5. Computing and Visualization components: Microarray Genome Mapper (developed by NYP), PCL and CS4 (developed by I2R)

Integrated Bio-Manufacturing Process using Goal Net

Manufacturing Diagnosis Kits

Initial State

System Initialized

Sample Collected Sample Processed DNA Sequencing Resutls

Similarity ResultsAnnotation Completed

Probe Designed

Experiment Generated

Data Nomalized

MicroarrayVisualized

MicroarrayDiscovered

Hits List Generated

Diagnosis KitsManufactured

Goal Achieved

New SampleObtained

New DataObtained

Filling ResultsObtained

Designed

Clinical Sample Prep Ready

Initialization

MicroarrayDiscovery

MicroarrayVisualization

MicroarrayData

Management

GenerateExperiment

ProbeDesign

SequenceAnnotation

HomologySearch withGridBlast

DNASequencing

SamplePrecessing

Viral/Bacterial/Pathogen

Isolate

ChromoWalking

GapFilling

DNASequencing

SampleProcessing

Finish

Diagnosis KitsManufacturing

Hits ListGeneration

Recloning/Amp

PrimerDesign

SequenceAnnotation

ClinicalSample

Prep

Probe ListPrep Ready

GenerateExperiment

ProbeList Prep

A Typical Bio-Manufacturing Workflow

Gap Filling

Recloning/ Amp

Chromo Walking

Viral/Bacterial/Pathogen isolate

Taverna KOOPlatform

DNA Sequencing

Sample Processing

Sequence Annotation

Probe Design

Primer Design

Homology Search w GRIDBLAST

Clinical Sample Prep

Generate Experiment

Microarray Data Mgt

Microarray Discovery Microarray Visualization

Hits List Generation

Diag Kits Manufacturing

Goalnet

Probe List Prep

Fig. 2 Architecture of the proposed system

Proposed Architecture for The Bio-Manufacturing Domain

Grid/web services

Grid/web services

Goal Net Process – 1Calling Kooplatform Plate Editor

Screen Shots of our System – 2Contig Result

Goal Net Process – 3Calling Taverna CompareXandY

Goal Net Process – 4Calling Kooplatform Array Normalization

Goal Net Process – 5Calling Kooplatform Class Discovery

Screen Shots of our System – 6Genome Result

Conclusion

• Mechanism for service and workflow orchestration

• Intelligent decision making according to the runtime situation

• Agent based integration and monitoring

• Design and development platform

Thank You!

top related