Modeling Protein Conformations by Guiding AlphaFold2 with Distance Distributions. Application to Double Electron Electron Resonance (DEER) Spectroscopy.

We describe a modified version of AlphaFold2 that incorporates experiential distance distributions into the network architecture for protein structure prediction. Harnessing the OpenFold platform, we fine-tuned AlphaFold2 on a small number of structurally dissimilar proteins to explicitly model distance distributions between spin labels determined from Double Electron-Electron Resonance (DEER) spectroscopy. We demonstrate the performance of the modified AlphaFold2, referred to as DEERFold, in switching the predicted conformations guided by experimental or simulated distance distributions. Remarkably, the intrinsic performance of AlphaFold2 substantially reduces the number and the accuracy of the widths of the distributions needed to drive conformational selection thereby increasing the experimental throughput. The blueprint of DEERFold can be generalized to other experimental methods where distance constraints can be represented by distributions. The training data, model weight, and prediction results for DEERFodl are provided as follows: DEERFold/ # Main project directory ├── data/ # Dataset directory │ ├── train.json # Training dataset │ └── val.json # Validation dataset └── model/ # DEERFold model weight └── DEERFold.pt # Model weights file DEERFold_benchmark/ # All prediction results shown in paper ├── general_tests/ # General benchmark tests │ ├── AK/ # Target │ │ ├── target/ # Target structure predictions │ │ └── unconstrained/ # Predictions without constraints │ ├── ASCT2/ │ ├── DgoT/ │ ├── LAT1/ │ ├── MCT1/ │ ├── RBP/ │ └── STP10/ ├── LmrP/ # LmrP predictions │ ├── Experiment1/ # Experiment1 set │ │ ├── Experiment1.csv # DEER distance distributions │ │ ├── full/ # Full MSA predictions │ │ └── neff5/ # Reduced MSA (Neff=5) │ ├── Experiment2/ # Experiment2 set │ ├── Simulation1/ # Simulation1 set │ ├── Simulation2/ # Simulation2 set │ └── Unconstrained/ # Predictions without constraints ├── PfMATE/ # PfMATE predictions │ ├── Experiment1/ # Similar as LmrP │ ├── Experiment2/ │ ├── Simulation1/ │ ├── Simulation2/ │ └── Unconstrained/ └── pgp_mouse/ # Pgp predictions ├── 7ZK4_sampling/ # Sampling results from simulation data of PDB:7ZK4 structure ├── 7ZK7_sampling/ # Sampling results from simulation data of PDB:7ZK7 structure ├── ADP-Vi/ # Predictions of experimental data from ADP-vanadate bound state │ ├── Experiment1/ │ └── Experiment2/ ├── Apo/ #Predictions of experimental data from Apo state └── Unconstrained/ # Predictions without constraints └── neff10/ # Reduced MSA (Neff=10)