A large amount of data is generated each year by the scientific community. These data must be collected with sufficient metadata and controls and have an adequate level of completeness and accuracy that allow it to be utilized and analyzed accurately. The data must also be stored in a machine readable format that allows it to be easily validated and loaded into a database. We are developing machine readable templates for data captured manually, which provides guidelines on capturing the data, metadata and available controls and defines the level completeness and accuracy required. We are also providing similar guidelines for data captured automatically by scientific equipment, such as genotyping systems, or data generated by analytical software. We have developed data templates for plant accession passports and genotyping data produced in work by the Generation Challenge Program (GCP, www.generationcp.org) and its partners, and are developing additional templates for mapping, QTL, SNP genotyping and plant phenotypic (evaluation) data in collaboration with GrainGenes (wheat.pw.usda.gov/), MaizeGDB (www.maizegdb.org/) and Gramene (www.gramene.org/). The templates are provided in both Excel and text formats, which are validated and converted to XML for storage. The software is generic enough to support a range of formats and can conform to most XML schemas. The XML form of the data is then transformed using XSL (Extensible Stylesheet Language) to the various formats required for database submission, visualization and analysis. The templates and software are available on the GCP bioinformatics portal (www.generationcp.org/bioinformatics). (Texte intégral)