Skip to main content

Table 3 A prediction performance comparison to show the effectiveness of incorporating neighbourhood information

From: A method to improve protein subcellular localization prediction by integrating various biological data sources

  No NI(*) NI(*)
Prediction coverage (%) 60 84
Measure-I (k = 3) (%) 86.14 87.50
Measure-II (%) 63.52 67.76
   Mitochondrion 35.12 57.85
   Vacuole 31.58 15.30
   Spindle pole 38.89 60.00
   Cell periphery 34.92 30.34
   Punctate composite 19.67 16.50
   Vacuolar membrane 8.11 22.73
   ER 53.02 51.89
   Nuclear periphery 50.00 48.98
   Endosome 40.74 44.12
   Bud neck 36.11 52.73
   Microtubule 45.46 31.58
   Golgi 23.81 42.86
   Late Golgi 21.74 29.03
   Peroxisome 33.33 65.00
   Actin 52.94 58.62
   Nucleolus 32.91 58.62
   Cytoplasm 79.88 80.68
   ER to Golgi 100.00 66.67
   Early Golgi 26.67 48.87
   Lipid particle 9.09 6.67
   Nucleus 77.91 80.59
   Bud 7.69 23.81
Measure-III (%) 39.07 45.15
  1. (*) Note: No NI: Prediction without incorporating neighbourhood information; NI: Prediction with neighbourhood information.