Secondary Structure Predictor help page

Secondary Structure Predictor, is a classic three-states predictor (alpha helix, beta strand, random coil) calibrated on soluble globular proteins. This class contains only one prediction method, that is based on Neural Network (NN) computation and on "profile" generation.

Secondary Structure Predictor (NN-based and Profile-based)

  • Target: The real prediction target is represented by the secondary structures (like alpha-helices and beta-strand) of globular proteins, more extensively of soluble proteins. These structures and their features are widespread and often relatable to all proteins that fold and reside in a water environment. This tool should not be used for the membrane proteins. Its results for this kind of proteins are unreliable and should not be taken in account.
  • Output: The NN output is represented by three predicted states for each aminoacidic position, each one representing the probability to be in an helix or in a strand or in a coiled state.
    The output is graphically presented in three colors, according to the legend at the bottom of the image, one for each predicted secondary-structure state. The difference between the two biggest predicted probabilities is also represented as a continuous line interpolating each discrete positional value.
    In the table, the results for each each segment is labelled accordingly to the predicted feature:
    • Alpha - for alpha-helices
    • Beta - for beta-strands
    • Other - for random coils