Get Started!

The exemplary output can be found here.


  • Step1: Select if input a PDB structure or a protein sequence
  • Step2: For structure input: Input the PDB code or upload a PDB formatted file;
    For structure in the PDB file, only residues with backbone N, CA and C are taken into account, only lines initial with "ATOM" or "HETATM" are taken into consideration! MSE residues will be transformed into MET. Too many atoms may hinder the APBS program and, thus, may not be returned.
    If you have any problem with the preparation of your file, please let contact us.
                   For sequence input: Input fasta formatted sequence or sequences, or upload a sequence file that contain all sequences in fasta format;
    Theoretically, RBscore can predict many sequences in a single job. However, we noticed that memory and job time are limited by our queue system. Ideally, within 5 sequences is OK for a single job.
  • Step3: is only for structure based prediction, to input Chain ID. Please note a chain ID could be either uppercase or lowercase.
  • Step4: Before submission, users can leave their email address to get prediction and notification by email. This is optional.
    In order to avoid melicious attack, we require users to validate with recaptcha code. If you need to run batches of jobs, please contact us.
    Finally, just click .
  • Advanced options: Users can define the specificity or sensitivity as expected. The value should always between 0 and 100 %.
    And users can choose the prediction model for the SVM approach. By default, the prediction is based on SVM model 628, which was built on 628 nucleic acid binding proteins.

  • Output

    Summary of prediction

    Jod ID-- an automatically assigned number to trace the job.
    Sequence length-- length of the protein.
    Possible binding sites No.--Number range of binding sites predicted from the protein sequence.
    Download results data--download the compressed file of all result data.
    Binding sites Number--Number of binding sites predicted. This is for binary prediction of binding sites.

    Predictions based on different features

    In this part, users can plot the prediction scores on protein surface and explicitly show the binding region.
    The plotted prediction score can be:
    1. RBscore
    2. SVM based prediction
    3. RBscore based on only electrostatics
    4. RBscore based on only sequendce conservation
    5. RBscore based on only solvation energy
    Further, users can load ligand, save molecule or save figures(as png or jpg format).

    Features map on protein structure

    In this part, users can plot the related features on protein surface to show the feature distribution.
    Features include: conservation entropy, which is similar to the result of ConSurf and electrostatics, which is same as result of APBS tools.
    Also, users can load ligand, save molecule or save figures(as png or jpg format).

    Deatailed Results

    RBscore-- total prediction score
    Qscore-- prediction based on ELECTROSTATICS
    CEscore-- Prediction based on CONSERVATION ENTROPY
    SOLscore-- Prediction based on SOLVATION ENERGY
    Entropy-- Conservation entropy
    RBpred-- Binary prediction by RBscore
    SVMscore-- prediction based on SVM
    SVMpred-- Binary prediction by SVM
    +: RNA binding
    -: non-RNA binding