Skip to main content

Table 2 Software performance evaluated using simulated transposition events in the Arabidopsis genome

From: TE-Tracker: systematic identification of transposition events through whole-genome resequencing

Software

RetroSeq

TE-Tracker

Delly

Hydra

GASVPro

Variation Hunter Common Law

Input data

PE reads

MP reads

PE reads

MP reads

PE reads

PE reads

Filter

None

>10 supporting pairs

>2 supporting pairs

>10 supporting pairs

>2 supporting pairs

>10 supporting pairs

Filtered predictions

190

351

795

10,366

6,448

26

FP

20

82

564

10,017

6,358

20

# Insertion found

146

260

282

139

247

6

# Insertion + correct donor found

128

244 (243)

214

139

225

0

Positive predictive value (PPV)

67.3%

69.5%

26.9%

1.3%

3.5%

0%

Sensitivity

42.6%

81.3%

71.3%

46.3%

75%

0%

  1. [ ] Insertion found at +/- 300 bp.
  2. [] Paired-end (PE) reads were generated using ART and mate-pair (MP) reads were generated using SimSeqG. If programs can deal with both types of input data, we chose to report only the results obtained from the sequencing protocol that led to the best metrics.
  3. A transposition event is qualified as found when at least one line in the output file has either one or the other side of a cluster overlapping the insertion site (for TE-Tracker, only the acceptor site is considered); A transposition event is qualified as found with donor when at least one line in the output file spans both the origin and destination sequence (for TE-Tracker the acceptor/donor nature of the site is taken into account). Even when the correct donor is identified for an insertion locus, other possible donors are often reported due to sequence similarity. For TE-Tracker, we display the number of cases where the donor-scoring feature distinguishes the real donor from all reported ones in parentheses. This feature is unique to TE-Tracker. The best detection statistic is displayed in bold in relevant rows.