DDG Upon Mutation(RDE)
1 Introduction
A flow-based generative model1 (named Rotamer Density Estimator, RDE) estimates the probability distribution of conformation , uses entropy as the measure of flexibility and predict the binding ∆∆G.
RDE1(ROTAMER DENSITY ESTIMATOR), a flow-based generative model that accurately estimates the probability distribution of protein side chain conformations through unsupervised learning of large amounts of protein structure data. The protein-protein binding energy change dataset SKEMPI22 is then used to train the RDE-Network, which is based on the idea that residues at the binding interface tend to become less flexible (i.e. entropy decreases) when proteins bind, and that this loss of entropy correlates with binding affinity. Therefore, by comparing the entropy loss of wild-type and mutant protein complexes, the effect of mutation on binding affinity can be estimated. In other words change in binding free energy ∆∆G can be predicted.
Figure 1. The overall architecture of RDE.
Figure 2. The performance of RDE.
2 Parameters
| Name | Description |
|---|---|
| PDB File | PDB file of protein complex structure. |
| Target Chains | Choose the protein chain of interest, it is recommended to choose a complete complex. |
| Mutation List | Enter a mutation string, for example, FP2Y indicates that position 2 on chain P has mutated from F to Y. If a variant has multiple mutations at the same time, separate them with commas, and each line represents a variant. |
3 Results Explanation
Output a result file in CSV format:
| Column name | Description |
|---|---|
| Mutstr | The mutation that occurred in this variant. |
| DDG Pred | The change in binding energy caused by this mutation (mutant minus wildtype), the smaller the value, the stronger the binding after the mutation, and the more important the mutation site is for binding between the receptor ligand. |
4 Reference
[1] Shitong Luo, Yufeng Su, Zuofan Wu, Chenpeng Su, Jian Peng, and Jianzhu Ma. Rotamer density estimator is an unsupervised learner of the effect of mutations on protein-protein interaction. bioRxiv, pp. 2023.02. 28.530137, 2023. https://doi.org/10.1101/2023.02.28.530137
[2] Justina Jankauskaitė, Brian Jiménez-García, Justas Dapkūnas, Juan Fernández-Recio, Iain H Moal, SKEMPI 2.0: an updated benchmark of changes in protein–protein binding energy, kinetics and thermodynamics upon mutation, Bioinformatics, Volume 35, Issue 3, February 2019, Pages 462–469. https://doi.org/10.1093/bioinformatics/bty635