Number of full-length transcripts identified by each analysis.

Sugarcane transcripts were first mapped against full-length grass transcripts and were then assigned as full-length if the CDS coverage was ≥95% and the identity was ≥80% (Analysis I). In this analysis, we found 9,960 full-length contigs, which aligned against 29,914 grass genes. Analysis II took into account more than one contig that mapped to the same grass gene (Figure S2), and non-overlapped coverage was calculated. Those that fit in the criteria of coverage ≥95% and identity ≥80% were considered full-length. This analysis yielded 26,384 contigs representing 3,952 unique grass genes and therefore 3,952 full-length sugarcane transcripts. For Analysis III, we calculated the average CDS size based on six grasses (Table S2), and all contigs with a predicted protein larger than this average CDS size were considered to be full-length.Number of full-length transcripts identified by each analysis.