Method at a Glance

DISPLAR is a neural network method. Given the structure of a protein known to bind DNA, the method predicts residues that contact DNA. The inputs to the neural network include position-specific sequence profiles and solvent accessibilities of each residue and its spatial neighbors. The neural network is trained on known structures of protein-DNA complexes. On our test set, DISPLAR shows prediction accuracy over 80% and coverage of over 60% of actual DNA-contacting residues.

Reference
  • Tjong , H. and Zhou, H.-X. DISPLAR: an accurate method for predicting DNA-binding sites on protein surfaces. Nucl. Acids Res. 35:1465-1477 (2007)


  •   Note: The size of protein can be submitted here is from 40 to 2000 residues.

    Predict DNA-binding site for your protein here

    If you have more than a few proteins and would like to run DISPLAR in batch mode, click here.

    • Type an arbitrary name for referencing your submission:
    • Enter your email address, where the prediction will be sent to:
    • Enter the chain(s) of your protein (do not include chains for DNA):
      No chain ID: leave underscore ("_") alone
      Single chain: delete "_" and type letter or number for chain ID, e.g., A
      Multiple chains: delete "_" and type chain IDs separated by comma without space, e.g., A,B,C
    • Upload your PDB file:

      OR paste your pdb (do NOT do both):

    • Note: DISPLAR is a 3D structure based predictor, not a sequence based.

    • Home

    • Input format

    • Output format

    • WESA

    • cons-PPISP

    • Download dataset


    • View Stats