Filtering and Validating Protein-Protein Docked Poses with Macromolecular Pose Filtering and MM-GBSA Residue Scanning

Tutorial Created with Software Release: 2026-2

Topics: Antibody Design, Biologics Drug Discovery, Structure Prediction & Target Enablement

Products Used: PIPER

This tutorial is written for use with a 3-button mouse with a scroll wheel.

Words found in the Glossary of Terms are shown like this: Workspacethe 3D display area in the center of the main window, where molecular structures are displayed

Tip: You can hover over a glossary term to display its definition. You can click on an image to expand it in the page.

Tutorial Content

1. Introduction to Validating Predicted Protein Complexes
2. Creating Projects and Importing Structures
3. Filtering the Docked Poses with the Macromolecular Pose Filtering Panel
4. Preparing the Filtered Poses
5. Validating the Filtered Poses with MM-GBSA Residue Scanning
6. Conclusion and References
7. Glossary of Terms

1. Introduction to Validating Predicted Protein Complexes

Accurate structural information is necessary for understanding biomolecular recognition and guiding rational design. While experimentally determined structures from X-ray crystallography, NMR, or cryo-EM provide the most reliable starting point for structural analysis, such data are not always available for every protein or protein-protein complex of interest. In many cases, computational structure prediction methods can provide useful alternatives, particularly when reliable homologous templates with similar sequences and folds exist. Advances in homology modeling, co-folding approaches, and machine learning-based predictors have significantly improved the reliability of predicted structures. However, predicting protein-protein complexes remains inherently more challenging than modeling individual proteins.

Given that experimental structures are not typically sought until much later in the lead optimization process, generating better complex models earlier can be highly valuable in driving forward a rational structure-based design approach. The most effective way to build confidence in a predicted model is to compare it against available experimental evidence or prior knowledge about the system. Experimental observations such as mutational binding data, cross-reactivity patterns, or known epitope and paratope residues can help determine whether the predicted interface is consistent with biological expectations. Similarly, experimental techniques such as hydrogen–deuterium exchange (HDX) can provide valuable constraints that support or challenge the predicted binding mode. Incorporating such information during model assessment helps identify biologically plausible conformations and ensures that subsequent computational analyses are based on structures that are consistent with known experimental behavior.

This tutorial uses the 2VXT structure, which is an antigen-bound murine antibody. In this tutorial, you will learn how to filter and validate an ensemble of antibody-antigen docked poses by leveraging data from alanine scanning. Here, 51 CDR residues of the 2VXT antibody were mutated site-by-site to alanine and the effect on target binding was measured. The assumption underlying this approach is that the residues that showed a significant loss in binding affinity are likely to play an important role in the interaction interface. Subsequently, you will map the binding poses with these interfacial residues and perform an in silico mutagenesis to alanine on these poses to validate whether the proposed binding modes are consistent with the observed mutational effects.

2. Creating Projects and Importing Structures

3. Filtering the Docked Poses with the Macromolecular Pose Filtering Panel

In the alanine scanning analysis performed on 51 CDR residues of the 2VXT antibody, a total of 11 residues were identified as critical for antigen recognition because their mutation to alanine resulted in significant loss of binding affinity, while the remaining 40 mutations were neutral with respect to binding affinity. The critical residues are distributed between the heavy chain and the light chain:

• Heavy chain residues: H: TYR 32, H: PHE 33, H: ASP 52, H: TYR 53, H: ASP 56, H: ARG 94, H: GLY 95, H: ARG 101
• Light chain residues: L: LYS 32, L: TYR 49, L: TYR 96

In this section, you will use the Macromolecular Pose Filtering panel to screen the 30 docked poses, ensuring that the filtered poses maintain contacts with the critical residues. Note that the wide geometric distribution of the mutations across both the heavy and the light chain is quite advantageous for filtering the docked poses. It spreads the ‘binding points’ across a larger area, thereby significantly limiting the number of possible orientations that can be satisfied.

The following interactive tutorial guides you through setting up and running the pose filtering job using specific residue constraints. You will first define the critical heavy and light chain CDR residues at the interface, the mutations to which were identified as deleterious. You will then specify the contact cutoff and the minimum number of residues that must match the filter. Finally, you will run the filtering job and view the filtered poses.

4. Preparing the Filtered Poses

In this section, you will reprepare the filtered poses using the Protein Preparation Workflow panel. It is important to note that the pKa of a particular titratable residue at the interface may differ between poses due to a change in its local environment. Therefore, to avoid calculating binding affinity changes from inconsistent wild-type states, you will modify the default settings to ensure that the wild-type state used for mutation is identical across all poses.

The following interactive tutorial guides you to set up and run the preparation job for the filtered poses with modified settings. To learn about the protein preparation workflow in detail, please consult the Introduction to Structure Preparation and Visualization tutorial.

5. Validating the Filtered Poses with MM-GBSA Residue Scanning

In this section, you will set up and run MM-GBSA residue scanning calculations on the filtered poses. The goal of running the MM-GBSA residue scanning calculations is to confirm consistency with experimental results: mutations predicted as deleterious by MM-GBSA should match those found to be deleterious in experiments, and likewise, neutral MM-GBSA mutations should align with experimentally neutral mutations that do not impact binding affinity.

Recall that of 51 CDR residues mutated to alanine, 11 mutations were found to be deleterious, causing a loss of binding, while the remaining 40 mutations were neutral with respect to binding affinity. For validation purposes, it is beneficial to incorporate as much experimental information as possible. Therefore, in addition to the deleterious mutations used for pose filtering, you will also test 6 neutral mutations, which are expected to have minimal impact on binding affinity when mutated to alanine. Neutral mutations can have no impact on binding affinity for one of two reasons: they have no impact on binding or that they are not at the interfaces. Since we do not know a priori whether these sites are interfacial or not, we do not include them in the initial filtering criteria; however, since we know their effect on binding we can use them as controls in the validation step.

To maximize the effectiveness of the mutational validation using MM-GBSA, you will select one residue per CDR loop to ensure broader geometric coverage of the binding interface:

• Heavy chain CDRs: H1 - SER 28, H2 - GLY 55, H3 - TYR 90
• Light chain CDRs: L1 - GLN 27, L2 - SER 52, L3 - SER 93

The following interactive tutorial guides you through running the MM-GBSA residue scanning calculations and analyzing the results for the first of the poses which passed the filtering.

Now that you have learned how to run an MM-GBSA Residue Scanning Calculation for one of the docked poses, follow the same steps to set up and run the calculations for pose 9 and pose 13. If you do not wish to run the job yourself, the pre-generated results for both the calculations are available in the tutorial archive. The results of the MM-GBSA ∆ Affinity predictions are summarised below:

6. Conclusion and References

In this tutorial, you learned how to use prior experimental knowledge to effectively filter and validate predicted protein-protein complex structures. In many practical protein-protein modeling projects, the challenge extends beyond predicting the correct binding orientation of two partners and includes identifying the biologically relevant conformations of the individual proteins themselves. In such cases, each binding partner may first need to be modeled independently using approaches such as AlphaFold, homology modeling, or related structure prediction methods. Because proteins can adopt multiple plausible conformations, several structures variants of one or both partners are often generated. Each possible combination of these structures can then be subjected to protein-protein docking, producing an ensemble of complexes. This substantially increases the size and diversity of the pose set and makes robust filtering strategies essential.

7. Glossary of Terms

Entries - a simplified view of the Project Table that allows you to perform basic operations such as selection and inclusion

Included - the entry is represented in the Workspace, the circle in the In column is blue

Project Table - displays the contents of a project and is also an interface for performing operations on selected entries, viewing properties, and organizing structures and data

Recent actions - This is a list of your recent actions, which you can use to reopen a panel, displayed below the Browse row. (Right-click to delete.)

Scratch Project - a temporary project in which work is not saved, closing a scratch project removes all current work and begins a new scratch project

Selected - (1) the atoms are chosen in the Workspace. These atoms are referred to as "the selection" or "the atom selection". Workspace operations are performed on the selected atoms. (2) The entry is chosen in the Entries table (and Project Table) and the row for the entry is highlighted. Project operations are performed on all selected entries

Working Directory - the location where files are saved

Workspace - the 3D display area in the center of the main window, where molecular structures are displayed