Conformational B-Cell Epitopes Prediction from Sequences Using Cost-Sensitive Ensemble Classifiers and Spatial Clustering

Procedure 2

Procedure Spatial clustering algorithm.

Begin

  Data preparation: 𝐶: the set of predicted antigenic determinant residues; 𝑇: threshold for spatial clustering; 𝑃: 3D structure

  (predicted or observed) of the protein.

  Calculate the max distance between any two residues in  𝐶.

  IF the max distance is greater than the pre-defined threshold  𝑇

   Clustering the residues in  𝐶  into two smaller clusters according to their spatial positions with standard  𝐾-means

   algorithm: 𝐶₁  and  𝐶₂

   EpitopeClusters1 = SpatialClustering(𝐶₁,𝑇,𝑃)

   EpitopeClusters2 = SpatialClustering(𝐶₂,𝑇,𝑃)

   EpitopeClusters = EpitopeClusters1  ∪  EpitopeClusters2

  ELSE

   EpitopeClusters = 𝐶

  END IF

  RETURN EpitopeClusters

End