International HapMap Project   Home | About the Project | Data | Publications | Tutorial

XML and XML Schema in the HapMap project

Introduction

Data are exchanged between the DCC, genotyping groups and analysis groups in a format based on the eXtensible Markup Language. More specifically, the data format is strictly defined in the XML Schema language, itself based on XML. The benefits of using XML and related technologies, especially XML Schema, for a project like the HapMap project are numerous, some which are listed here:

• Extensibility: it is very easy, compared to other formats, to extend XML and/or transform it into other formats. For instance, it would be easy work creating an eXtensible Stylesheet Language (XSL) transformation stylesheet that would transform a set of HapMap haplotype blocks from a chromsome region (result of a database query perhaps) into BSML or AGAVE XML-formats or similar, or even non-XML formats if desired.
• Enforcing data formats: in our experience, having a common file format that is strictly adhered to is worth its weight in gold. Using the older DTD method of defining an XML-format is useful despite its limitations and arcane syntax, but with the advent of XML Schema, one can define database-oriented content much more strictly in terms of datatypes and other such important database-concepts. It will make a huge difference for HapMap data handling and software development to be able to e.g. check if a submission datafile is properly formatted by simple running it through a validating parser (many of those available).
• Easy programmatic access: there is a wealth of software in various programming languages available for processing XML, both commercial and free. That frees the developer from having to actually write parsers for file formats and enables him/her to concentrate on the structure of the document in question, via a variety of content models (DOM-tree, SAX-events)

Which schema definition to use?

Every XML document that goes between the DCC and other HapMap collaborators must A) refer to a specific URL on the Internet where the schema is located, and B) contain XML content that validates against that schema. See this directory for sample XML-documents. You can browse the HapMap schema by opening the schemaLocation URL at the top of each of the sample XMLs (http://www.hapmap.org/xml-schema/2003-11-04/hapmap.xsd as of May'04), in a web browser. There are documentation snippets attached to each non-trivial element defined in the schema.

As the HapMap schema will change over time (despite our best intentions to get it right the first time!), there will unavoidably be several versions of the schema referenced in XML documents from various points in time. All versions will be available through our webserver for validating against if necessary (see list of all versions to date here), but there will always be ONE primary schema that should be used when the groups need to prepare data for submission to the DCC at any given time.

Schema Layout

The HapMap schema consists of several small files that each describe a single kind of large-ish object (a SNP, a protocol), several related smaller ones or common datatypes or element groups. These components are all tied together with the master file, hapmap.xsd, that includes all the others. See the schema files themselves in this dir for descriptions of the various elements and datatypes(as XML <annotation> elements).

Last updated : index.html,v 1.2 2004/05/06 15:00:11 mummi Exp