Copyright information:Taken from "Ensembl 2006"Nucleic Acids Research 2005;34(Database issue):D556-D561.Published online 28 Dec 2005PMCID:PMC1347495.© The Author 2006. Published by Oxford University Press. All rights reserved The column legends indicate the species, reference dataset and assembly release number. UniSw indicates the Swiss-Prot (curated) part of UniProt. RefSeq indicates the curated part of RefSeq (i.e. excluding XP entries). Identical trends are seen in all four comparisons of human and mouse against UniSw and RefSeq. The four colours indicate the quality of the match to the reference dataset: blue indicates an exact match; maroon indicates matched ends with some internal mismatch/indel; yellow indicates an incomplete match and green indicates reference sequences that are missing from the gene build. There are multiple reasons for this improvement, including improvements in assembly quality, cDNA resources and algorithmic improvements to the gene build.