Consensus-based Inverse Ligand binding predictor (ILbind)

ILbind standalone

ILbind is a consensus predictor that combines the two complementary inverse ligand binding predictors implemented using FINDSITE and SMAP and Support Vector Machines. The program is available for all non-commercial users here. Installation and usage instruction is available here. The program may only be installed on computers with Linux OS. If user prefers not to install the program a web service is available.

ILbind webserver

Please follow the three steps below to make predictions:

1. Enter protein id(s) followed by predictions from FINDSITE and SMAP

Please enter each protein in a new line in following format:

"ID, alignment length predicted by FINDSITE, raw score predicted by SMAP"

Up to 1000 proteins allowed. Protein id will be trimmed to 8 characters. Please consult help to learn how to obtain FINDSITE and SMAP predictions.

2. Provide your e-mail address (required)

Please provide your e-mail address to be notified when results are ready.

3. Predict:

Click button to launch prediction.


Upon the usage the users are requested to input predictions from FINDSITE and SMAP methods. Beside the target protein and ligand users should also have template with known ligand-protein complex, which is used to obtain those predictions by following these steps:

  1. Download and install FINDSITE package according to instructions available on this website.
  2. Download and install SMAP package according to instructions available on this website.
  3. Run FINDSITE, using the target protein and the chosen templates as FINDSITE's threading library.
  4. Run SMAP using the target protein and each of the chosen templates.
  5. Collect maximal length of aligned sequence from FINDSITE's result files, and maximal raw score from SMAP's result files
  6. Input predictions and email address into text fields above, and click Run ILbind

More detailed instruction on how to obtain SMAP and FIndsite predicitons is available here: ILbind.pdf


Each line in the datasets files include two columns, where the first column identifies if the protein is binding (1) or non-binding (0), and the second column is the PDB id.


We acknowledge with thanks the following software used as a part of this server: