A FRAMEWORK BASED ON WEB SERVICES ORCHESTRATION FOR BIOINFORMATICS WORKFLOW

MANAGEMENT

Laboratory for Bioinformatics (LBI), Institute of Computing (IC) - UNICAMP

Luciano Antonio Digiampietri

João Carlos Setubal

Cláudia Bauzer Medeiros

PhD Student:

Advisor:

Co-advisor:

Motivation

Genome assembly and annotation pipeline

assemblyreads contigs human

validation

incrementalassembly

new reads generation

assembly OK annotation

Motivation

The growth of bioinformatics activities– Data– Services

Data and services don’t use public standards

Goals

Specification and development of a framework that allows:– Data integration;– Service integration;– Modeling of complex tasks as workflows;– Coordination of workflow execution

This talk

Work in progress– no results yet

Overview: user interaction

Related issues

Web services– “a software application identified by a URI,

whose interfaces and bindings are capable of being defined, described, and discovered as XML artifacts” [W3C:webservices]

Related issues

Workflows– Workflows represent a set of activities

to be executed, their interdependencies relations, inputs and outputs.

activity1

activity2

activity3input data1

input data2

input data3

activity4

output data

Related issues

Service coordination– Service orchestration is a centralized

mechanism that describes how diverse services can interact. This interaction includes message exchange, business logic and order of execution;

– We are using PBEL4WS as the specification language for service orchestration

Related issues

Bioinformatics tools and data– Selection of basic bioinformatics tools for

genomic assembly and annotation;– Selection of some important data sources;– Use of tools and data ontology.

Related issues

Example of part of a tool ontology:

Alignment service

Local Alignment Global Alignment

Heuristic Alignment

Non-Heuristic Alignment

Non-Heuristic Alignment

Heuristic Alignment

The framework

Workflow management

Service layer

Service discovery Service request

application1 application2

Workflow discovery

Workflow repository

Workflow design Workflow engine

User interface

Service Catalog application3

The framework

Service layer – bioinformatics basic Web services

• assembly,• matching,• consensus, • etc

Service catalog layer – stores Web services'

• syntactical description;• semantical description;• URI.

The framework

Service discovery – search by:

• functionality,• context,• syntax.

Service request layer – management of each Web service solicitation:

• Sending input data;• Receiving results;• Detecting service failure.

The framework

Workflow engine layer – controls execution of all workflow tasks, via orchestration. – The main functions:

• interpretation of the process (or task) definition,• creation and management of process instances,• navigation between activities,• supervisory functions.

Workflow design layer – supports workflow specification and edition. – The facilities provided are:

• graphical interface for workflow edition,• service list,• interface description of selected services,• syntactical check of workflow.

Methodology

A bottom-up approach:

Service1

Service2

Service3

Workflow design and

engine

Workflow management

Service discovery

and requisition

Methodology

Services– bioinformatics basic services

• specification• development.

– metadata types definition Service discovery

– development of techniques for service discovery and request using syntactic and semantic search mechanisms.

Methodology

Workflow– specification and development of methods

for workflow design and execution;– Design of workflows is being done using

WOODS*; – Specification and implementation of an

orchestration mechanism.

*Seffino, L.A., Medeiros, C.B., Rocha, J.V.R., Yi, B.:WOODS- a spatial decision support system based on workflows. Decision Support Systems 27 (1999) 105-123

Conclusions

The main contribution is the framework itself:– It allows multi-institutional cooperation,

sharing:• Data• Tools• Workflows

– It can be the interface among various kinds of users of different research centers

Conclusions

Other contributions lie in– Scientific workflows specification and

publishing (using Web services as basic units);

– Semantic specification of bioinformatics tasks;

– Definition of a generic methodology for data and tools integration.

Thank you!Laboratory for Bioinformatics www.lbi.ic.unicamp.br

Institute of Computation (IC) www.ic.unicamp.br

University of Campinas (UNICAMP) www.unicamp.br

Luciano Antonio Digiampietri luciano@ic.unicamp.br