Production of simulated benchmark transcriptomes and read sets

Simulated benchmark transcriptomes and corresponding RNAseq read sets were produced from A. thaliana and humans as follows. A modified Flux Simulator v1.2.1 pipeline [42] illustrated in the Additional file 1: Pipeline 1 was used to produce ~ 250 M 76 bp read pairs for each dataset. The No0-NoAlt transcriptome was generated based on the A. thaliana No-0 genomic sequence and no alternative splice events were included. The Col0-Alt transcriptome was generated based on the A. thaliana Col-0 genomic sequence and the version 3 atRTD transcriptome model. The expressed transcriptome in this dataset contains zero to nine alternative splice events (one to ten isoforms) per gene where only the isoforms unique at the protein level were included. The transcriptome for the Human dataset was based on the HG38 reference genome and transcriptome. It contains zero to 14 alternative splice events (1 to 15 isoforms). Only reference transcripts with full coverage of RNAseq data (all positions are required to be covered by at least one read) were included in the benchmark datasets, as transcripts without full coverage cannot be correctly assembled as a single contig. Additional file 2: Table S1 shows the number and distribution of isoforms per transcript for each dataset. All benchmark datasets used in this study are available from: http://bioinfolab.unl.edu/emlab/consemble/