Fig. 4

FINDER versus other pipelines on different groups of genes in three model species—a A. thaliana, b O. sativa, c Z. mays. F1 score is the harmonic mean between sensitivity and specificity. Higher F1 score indicates better agreement with the reference transcript models. We created groups of transcripts that have similar characteristics as shown in the y-axis legend. A pool of transcripts was created containing multi-exonic transcript predictions, from each pipeline, that has a complete intron chain match with at least one reference annotation. Mono exonic transcripts were considered if at least 80% of the nucleotides overlap with one reference annotation. Transcript F1 scores, for each of the annotation pipelines, have been plotted as a bar graph. Even though all annotation pipelines are designed to serve the same purpose of annotating genomes, each pipeline adopts a different strategy. Each strategy has its own merits and demerits that lead to better annotation of a certain category of genes. This plot helps understand the performance of each annotation pipeline on different categories. The symbol “#” denotes the best annotator in each gene group. (Generated using ggplot2 v3.3.3)