Strategy for characterizing unannotated transcripts. (a) Novel transcripts predicted on the basis of the piling-up of mapped short reads by using the Cufflinks program. Unannotated transcripts were selected by checking overlap with Phytozome annotation (Sbicolor_79). To predict the functions of the unannotated transcripts, BLASTx searches were performed against known protein datasets, namely Uniprot (Rel. 2011_01) and RefSeq (release 45). (b) Differential expression of Cufflinks-predicted transcripts. The graph indicates the coverage of mapped reads from mock- (mock; green), or pathogen-infected (infected; purple) leaves, in GBrowse. Exon models were predicted by the piling-up of reads by the Cufflinks program (Novel Cufflinks-predicted gene; blue box). Gene models annotated in Phytozome are also shown (Phytozome annotation; green boxes). CUFF115357.1 was unannotated and was expressed differentially.