The GENCODE gene set has developed substantially between releases 3c and 7 (see Fig. 4). Release 3c was the first complete merge set containing all the CCDS transcripts and used by the 1000 Genomes Consortium as its reference annotation. GENCODE release 7 is the reference for the analysis of ENCODE project data carried out in 2011. First-pass manual annotation has been done on 18 chromosomes (chr), and HAVANA still has chr14–19 to complete before the whole genome has been fully manually annotated. Supplemental Figure 4 demonstrates how the number of lncRNAs has increased dramatically with the full manual annotation of the chromosomes. The number of protein-coding loci has decreased significantly between GENCODE releases 3c and 7; however, this is almost entirely due to the removal of poorly supported automatic annotation models, particularly between releases 3c and 3d, where 1004 models were removed from the automatic annotation set. All GENCODE small noncoding RNAs are Level 3 and, as such, show a different pattern to other locus biotypes with their numbers dropping between 3d and 4 as incorrect automated gene models are removed and remaining stable thereafter.
