Biological data retrieval from web resources often necessitates multi-step access to several information sources. Web services are being developed to interact in a unified manner with heterogeneous databases. Composition of web services is then required to chain several elementary queries. However, web services are not yet available for any data sources thus limiting the scope of such strategy. The X-collect generic application has been designed to execute user-defined multi-step scenarios. Demos can be scheduled at the workshop. Encapsulation into web services makes such user scenarios re-usable and available for invocation by any application. Demos can be scheduled at the workshop. Work in progress deals with metadata registries to facilitate databases and/or web services discovery.