PPIformer Web (CPU version)

Computational Design of Protein-Protein Interactions

Image

PPIformer is a state-of-the-art predictor of the effects of mutations on protein-protein interactions (PPIs), as quantified by the binding free energy changes (ddG). PPIformer was shown to successfully identify known favourable mutations of the staphylokinase thrombolytics and a human antibody against the SARS-CoV-2 spike protein. The model was pre-trained on the PPIRef dataset via a coarse-grained structural masked modeling and fine-tuned on the SKEMPI v2.0 dataset via log odds. Please see more details in our ICLR 2024 paper.

Inputs. To use PPIformer on your data, please specify the PPI structure (PDB code or .pdb file), interacting proteins of interest (chain codes in the file) and mutations (semicolon-separated list or file with mutations in the standard format: wild-type residue, chain, residue number, mutant residue). For inspiration, you can use one of the examples below: click on one of the rows to pre-fill the inputs. After specifying the inputs, press the button to predict the effects of mutations on the PPI. Currently the model runs on CPU, so the predictions may take a few minutes.

Outputs. After making a prediction with the model, you will see binding free energy changes for each mutation (ddG values in kcal/mol). A more negative value indicates an improvement in affinity, whereas a more positive value means a reduction in affinity. Below you will also see a 3D visualization of the PPI with wild types of mutated residues highlighted in red. The visualization additionally shows the attention coefficients of the model for the nearest neighboring residues, which quantifies the contribution of the residues to the predicted ddG value. The brighter and thicker a residue is, the more attention the model paid to it.

PPI structure

PDB code

Protein Data Bank identifier for the structure (https://www.rcsb.org/)

Partners

Protein chain identifiers in the PDB file forming the PPI interface (two or more)

Or .pdb file instead of PDB code (your structure will only be used for this prediction and not stored anywhere)

Mutations

List of (multi-point) mutations

SC16A;FC47A;SC16A,FC47A for three mutations: serine to alanine at position 16 in chain C, phenylalanine to alanine at position 47 in chain C, and their double-point combination

Or file with mutations

Examples (click on a line to pre-fill the inputs)

PDB code	Partners	List of (multi-point) mutations

Predictions

Download predictions as .csv

About this web

Use cases. The predictor can be used in: (i) Drug Discovery for the development of novel drugs and vaccines for various diseases such as cancer, neurodegenerative disorders, and infectious diseases, (ii) Biotechnological Applications to develop new biocatalysts for biofuels, industrial chemicals, and pharmaceuticals (iii) Therapeutic Protein Design to develop therapeutic proteins with enhanced stability, specificity, and efficacy, and (iv) Mechanistic Studies to gain insights into fundamental biological processes, such as signal transduction, gene regulation, and immune response.

Acknowledgement. Please, use the following citation to acknowledge the use of our service. The web server is provided free of charge for non-commercial use.

Bushuiev, Anton, Roman Bushuiev, Petr Kouba, Anatolii Filkin, Marketa Gabrielova, Michal Gabriel, Jiri Sedlar, Tomas Pluskal, Jiri Damborsky, Stanislav Mazurenko, Josef Sivic. "Learning to design protein-protein interactions with enhanced generalization". The Twelfth International Conference on Learning Representations (ICLR 2024). https://arxiv.org/abs/2310.18515.

Contact. Please share your feedback or report any bugs through GitHub Issues, or feel free to contact us directly at anton.bushuiev@cvut.cz.

Image