NMDtxDB: data-driven identification and annotation of human NMD target transcripts

TABLE 5.

Benchmark of sample 33G-30 versus CHESS

Feature Parameter set Recall Precision F1
Transcript Conservative 23.6 49.6 32.0
Transcript conservative_trimming 23.6 49.4 31.9
Transcript Default 24.2 46.0 31.7
Transcript default_noguide 9.7 38.9 15.5
Transcript Mix 22.7 52.7 31.7
Transcript mix_noguide 11.2 55.7 18.7
Exon conservative 49.7 77.0 60.4
Exon conservative_trimming 49.8 76.9 60.5
Exon Default 50.4 75.0 60.3
Exon default_noguide 39.0 79.8 52.4
Exon Mix 49.4 80.3 61.1
Exon mix_noguide 40.0 86.8 54.8
Intron conservative 49.8 88.7 63.8
Intron conservative_trimming 49.9 88.7 63.9
Intron Default 51.1 86.0 64.1
Intron default_noguide 47.3 87.4 61.4
Intron Mix 51.2 93.4 66.1
Intron mix_noguide 46.2 95.4 62.3
  • Parameter sets: default: StringTie with default parameters. noguide: StringTie run without reference. Conservative: uses the ‐‐conservative flag (same as -t -c 1.5 -f 0.05). trimming: enables trimming (-t flag). mix: uses mix for combining Illumina and Nanopore reads. Bold entries represent the top metric for each feature group.

This Article

  1. RNA 30: 1277-1291