Fig. 5From: Cleaning by clustering: methodology for addressing data quality issues in biomedical metadataThree clusters for the key “strain”, see panels (a), (b) and (c), respectivelyBack to article page