Our offer is based on SAFAN-ISP (In Silico Profiling), our proprietary technology that facilitates:
- finding new targets for your molecule
- finding potential off-targets
- elucidating the biological mechanism of your compound
- planning your future experiments.
Our fragment based predictions allow you to:
- patent new compounds
- get information comparable to lab experiments at lower cost.
Starting from the structure of a list of compounds or peptides we deliver the binding affinities for protein targets.
- We connect the specific interaction with:
- Disease database for repositioning opportunities
- Side Effect database for toxicology predictions
Here you can find an example of the computational profiling output.
Here is a slides deck describing our approach and its validation.
We have an ongoing partnership with Humana Biosciences to perform experimental validation of our computational results.
In the Virtual Screening process a library of compounds is screened against one single protein target to find the compounds most likely to binds with it.
SMALL MOLECULES DATABASE
The profiling process starts from a single small molecule or peptide and screens a protein target library to find the protein that most likely will bind it
PROTEIN TARGETS DATABASE
Our platform comes in two specialized variations: SAFAN-ISPSM for small molecules and SAFAN-ISPPEPT for peptides. They quickly and efficiently profile their compound, calculating the binding affinities between each ligand and more than 4500 targets from 15 different protein classes. They rely on our proprietary algorithm to forecast, affinities through fragments weight assignment using a ligand-based approach. We work with a refactored bioactivity database derived from the CHEMBL25 database.
- You can get a selection of SAFAN-ISPSM profiles ready to download concerning:
- 9488 from DrugBank
- 26510 from FooDB
- 75000 Natural
- Or we can profile your proprietary small molecule libraries.
Peptide-protein interactions play a critical role in the protein-protein interaction network with significant involvement in signal transduction and regulation. Many of these interactions are promising candidates as new leads for drug targets. However,
- Peptides often lack a distinct fold
- In many cases there is no data regarding the peptide binding site and/or the peptide backbone conformation.
- Structure-based modeling of these interactions is very challenging.
Using a ligand-based approach, SAFAN-ISPPEPT overcomes these challenges, to predict protein:peptides binding affinity quantitatively.
Our Technology Validation
SAFAN-ISPSM and SAFAN-ISPPEPT results were validated by the leave-one-out method, resulting in a Pearson correlation with experimental data higher than 0.9 and 0.8 respectively.
Pearson Correlation 0.91
Pearson Correlation 0.87
pChEMBL is defined as: -Log(molar IC50, XC50, EC50, AC50, Ki, Kd or Potency)
Expected binding predictions (pChEMBL) with error <1
Here you can see the expected error according to the known percentage of the small molecule. For instance, if we know only 40% of the input molecule, we can expect 84% of the predictions to have an error of less than 1.
Here you can see the expected error according to the known percentage of the peptide. For instance, if we know only 40% of the input peptide, we can expect 76% of the predictions to have an error less than 1.
Our expertise based on many years of experience in basic and applied research in chemoinformatics and structural bioinformatics can help you succeeding in your research projects – including difficult homology modeling projects.
- Help in difficult homology modeling projects.
- Analysis of the dynamic properties of proteins and nucleic acids by molecular dynamic simulations.