Accurate fusion transcript identification from long- and short-read isoform sequencing at bulk or single-cell resolution

(Downloading may take up to 30 seconds. If the slide opens in your browser, select File -> Save As to save it.)

Click on image to view larger version.

Figure 4.
Figure 4.

Detection of fusion transcripts from MAS-ISO-seq of nine cancer cell lines. (A) Counts of fusion predictions according to cell line, prediction method, and benchmarking class assignment requiring a minimum of three long reads as supporting evidence and a minimum of two different methods agreeing on fusion predictions (referred to as the example proxy truth set). Counts of true positives (TPs), FPs, and false negatives (FNs) are shown based on this example proxy truth set. Total predicted fusions with minimum three reads support = TP + FP. (B) Numbers of MAS-ISO-seq reads identified as evidence for COSMIC fusions according to method. (C) Fusion transcript detection accuracy according to minimum long reads supporting evidence, and (D) precision versus recall plotted for methods all according to the example proxy truth set. Related plots based on each of the proxy truth sets evaluated are available through Supplemental File 2. (E) Rankings according to top benchmarking AUC for each method according to differently defined truth sets based on minimum fusion evidence read support followed by a minimum number of agreeing methods. (F) Precision–recall AUC for benchmarking methods according to truth sets based on Illumina short-read-supported fusion transcripts.

This Article

  1. Genome Res. 35: 967-986

Preprint Server