Figure 1

PET-to-gene annotation. PET-to-gene annotation is based on tag-to-exon annotation, through which the mapping locations of the 5' and 3' tags of each PET1 ditags (PETs with unique targets) are compared with exon locations of the knownGene, refSeq and MGC genes of UCSC mm5 database. To absorb the variations that may be present, exon boundaries are extended 100 bp (for internal boundaries) or 1 Kb (for external boundaries). Under such definition, both 'a' and 'b' (5' tags) match to exon 1, and both transcripts are denoted as S1E3T3; 'c' (5' tag) matches to intron 1, and the transcript is denoted as S1-2E3T3; while 'd' (3' tag) matches to exon 3, and the transcript is denoted as S2E3T3. All these PETs are annotated to the same gene. S, starting exon; E, ending exon; T, total exon.