Introduction to the Matrix

Colored squares in the matrix represent predictions in which at least one pair of interfacial residues is closer than 5Å and meets minimum confidence criteria (PAE < 15 and pLDDT > 50).

Matrix colors can be set to represent the SPOC score (default) or other metrics. The darker the color, the higher the confidence of the interaction.

Grey squares represent protein pairs where no residues meet the above interaction criteria.

White squares represent protein pairs where no prediction was made, usually because the proteins exceed ~3600 amino acids.


To examine only high-confidence predictions, click on the "Default" filter (SPOC > 0.75).

To add a custom filter, click "add," choose a metric from the drop-down menu, and set a cutoff value.

Rows, Columns, and Lists

To display a subset of proteins in the matrix, use "row" and "column" menus.

To zoom in on a region of the matrix, click and drag over the area of interest.

To view all of a protein's predicted binding partners as a list, click its name on the X or Y axis in the matrix or enter its name in the search window and press return.

For more information on a predicted PPI, click on it in the matrix or list, or enter the two proteins in the search window and press return.

Structure Viewer

Predictions generated by each of the three AF-M models can be viewed separately by clicking on "view." To download a structure prediction as a .PDB file, click the arrow in the structure viewer.

Protein pairs for which there is at least one structure in the PDB are marked by an orange dot in the matrix (even if the structure does not match the prediction). The corresponding PDB files are listed above the structure viewer and in the list view.

To compare predicted and experimentally-determined PDB structures, click on "superimpose PDB." Superposition is based on best match of interfacial residues.

To hide disordered regions of proteins in the structure viewer, filter by pLDDT > 50.

PAE Plots

To getter a better view of the PAE values of a specific amino acid pair, hover over the PAE plots.

pLDDT Plot

To toggle between pLDDT scores of the two proteins, click "view."

To detect unstructured segments of proteins that are predicted to become ordered when folded with another protein, look for large increases in pLDDT score between the models and baseline (the pLDDT value for the protein in the presence of a non-interacting partner). A good example is CDT1 in the CDT1-PCNA complex.

MSA Plot

To view the evolutionary rate of change of individual residues (low change, high conservation), click and drag over the region of interest. Refresh page to reset.


This table lists all predicted inter-residue contacts between the two proteins. The list was generated from structures that have not been "relaxed." Therefore, before undertaking mutagenesis studies, we recommend downloading structures of interest, relaxing them using AMBER, and then identifying interacting residues using PISA.