Skip to main content
. 2024 Jun 7;21(7):1349–1363. doi: 10.1038/s41592-024-02298-3

Fig. 2. Evaluation of transcript identification with a reference annotation for Challenge 1.

Fig. 2

a, Percentage of transcript models fully supported at 5′ ends either by reference annotation or same-sample CAGE data (left), 3′ end either by reference annotation or same-sample QuantSeq data (middle) and splice junctions (SJ) by short-read coverage or a canonical site (right). b, Agreement in transcript detection as a function of the number of detecting pipelines, c, Performance of tools based on spliced-short (top) and unspliced long SIRVs (bottom). d, Performance of tools based on simulated data. e, Performance of tools on known and novel transcripts of 50 genes manually annotated by GENCODE. f, Summary of performance metrics of tools for the cDNA-PacBio and cDNA-ONT benchmarking datasets. The color scale represents the performance value ranging from worse (dark blue) to better (light yellow). The graphic symbol indicates the ranking position of the tool for the metric represented in each row. LO, long (reads) only; LS, long and short (reads); Sen_kn, sensitivity for known transcripts; Pre_kn, precision for known transcripts; Sen_no, sensitivity for novel transcripts; Pre_no, precision for novel transcripts; 1/Red, inverse of redundancy.