TCGA exon annotation in GFF format from June, 2011

TCGA exon and isoform annotation in GFF format from June, 2011, using the hg19 assembly, correcting for various annotation errors. and clustering transcripts the following way: 1. cluster transcripts according to genome overlap in the same strand. 2. group transcripts if they have a shared splice-site. 3. select as gene id the symbol if there is one, or use "locus_#" if they don't have gene symbol, or use two symbols if they happen to overlap and share a splice-site.