Skip to main content

Advertisement

Table 3 Ten FEATURE predictions confirmed by direct experimental evidence.

From: Prediction of calcium-binding sites by combining loop-modeling with machine learning

   Loop   FEATURE score
Protein name Protein ID Chain ID Start End Sequence SiteDis (Å) Gap Loop
VITAMIN B12 RECEPTOR 1NQG_A 231 241 AYYSPGSPLLD 2.06 42.41 91.49
BEAN LECTIN-LIKE INHIBITOR 1DHK_B 89 96 VQPE SKGD 3.40 27.35 61.10
  1. Two predictions of which calcium binding have been observed experimentally near the predicted sites. Column 2 lists protein name that is PDB code followed by the chain identifier. We build structures for gaps of which 3D structures are not presented in the original PDB files. Information of these structures, referred as "loop", is given in column 3-5. Column 6 lists the distances between the predicted sites and the experimentally observed sites (SiteDis). Column 7 and 8 list FEATURE scores at the predicted sites in the structures where loop structures are missing (Gap) and the structures of which loop structures were rebuilt by modeling methods (Loop). These two are holo structures, but their calcium-binding sites are not completely structured. FEATURE is able to identify these sites in the presence of the rebuilt loop structures.