The development of a database for metabolomics -looking back on ten years of experience

摘要

Metabolome profiles of ca. 750 compounds obtained from blood samples from 28 day rat studies (OECD 407) were combined with toxicity profiles in one database over ten years to predict toxicity of new compounds. We provide detailed descriptions of procedures and recommendations for 'omics data-bases. Control of variability (biological, sampling/storage and technical measurement) is essential. At the start of large scaleprojects control variability should be extensively investigated. Reference (positivecontrol) substances should be used to evaluate and obtain a good signaloise ratio. Procedures should be documented in standard operating procedures and followed meticulously. Exact repeats of positive controls should be regularly performed, to assessvariability of positive responses. Control data should be regularly checked for shifts and analysed to obtain information concerningnormality. If possible, data should be analysed by multiple procedures and conclusions should be drawn based on a joint assessment, not unlike peer review processes in histopathology.