A multi-scale structural and biophysical atlas of TCR-peptide-HLA recognition dynamics

DynaTPH Overview DynaTPH is a multi-scale structural and biophysical dataset designed to support the investigation of T cell receptor-peptide-human leukocyte antigen (TCR-pHLA) complexes. The dataset integrates static experimentally determined TCR-pHLA complex structures, molecular dynamics (MD) simulation trajectories and corresponding structural frames, and biophysical descriptors. DynaTPH contains 256 curated TCR-pHLA complexes, including 205 HLA class I and 51 HLA class II complexes. For each complex, three independent MD replicas were performed with a production simulation length of 50 ns per replica. The resulting dataset contains 768 MD trajectories, corresponding to 38.4 μs of total simulation time and 768,768 trajectory frames. Dataset at a Glance Item Description TCR-pHLA complexes 256 HLA class I complexes 205 HLA class II complexes 51 MD replicas 3 per complex Production simulation 50 ns per replica Total MD trajectories 768 Total simulation time 38.4 μs Frames per trajectory 1,001 Total trajectory frames 768,768 Main data components Static Data, Dynamic Data, Feature Data, descriptor.csv, and rmsf.csv Dataset Construction DynaTPH was constructed from experimentally determined TCR-pHLA complex structures through structure curation and standardized preprocessing. Three independent molecular dynamics simulations were subsequently performed for each complex, followed by extraction of structural and biophysical features from the resulting trajectories. The resulting dataset is organized into five complementary data components: Static Data — curated and standardized TCR-pHLA structural data; Dynamic Data — molecular dynamics trajectories and frames generated for the 256 complexes; Feature Data — structural and biophysical features derived from the structures and MD simulations; descriptor.csv— structure-level descriptors for the TCR-pHLA complexes; rmsf.csv— a residue-level table containing RMSF values derived from the MD trajectories. Dataset Organization The DynaTPH dataset is organized as follows: DynaTPH/ ├── Static_Data.tar.gz ├── Feature_Data.tar.gz ├── descriptor.csv ├── rmsf.csv ├── Dynamic_Data_shard01.tar ├── Dynamic_Data_shard02.tar ├── ... └── Dynamic_Data_shard64.tar 1.Static Data Static_Data.tar.gz contains the curated and standardized static structural data for the 256 TCR-pHLA complexes. Each structure is provided in PDB format and named according to its PDB ID. Static Data/ ├── 1BD2.pdb ├── XXXX.pdb └── ... 2.Dynamic Data The Dynamic Data contain molecular dynamics trajectories and corresponding frames generated for all 256 TCR-pHLA complexes. Each complex is represented by three independent 50-ns MD replicas, resulting in 768 trajectories and 38.4 μs of total simulation time. Each trajectory contains 1,001 frames, corresponding to snapshots from 0 to 50,000 ps sampled at 50-ps intervals. The dynamic data are organized by PDB ID and simulation replica: Dynamic Data/ └── PDB_ID/ ├── Replica0/ │ ├── *.xtc │ └── Frames/ │ ├── *.pdb │ ├── *.pdb │ └── ... ├── Replica1/ │ ├── *.xtc │ └── Frames/ └── Replica2/ ├── *.xtc └── Frames/ Due to the large storage size of the these data, the Dynamic Data are divided into 64 archive shards for storage and transfer. Each shard contains the dynamic data for four TCR-pHLA complexes. The complete Dynamic Data collection consists of: Dynamic_Data_shard01.tar ... Dynamic_Data_shard64.tar The 64 shards collectively constitute the complete Dynamic Data component of DynaTPH. 3.Feature Data Feature_Data.tar.gz contains structural and biophysical features derived from the curated structures and molecular dynamics trajectories. These features include descriptors such as RMSD, RMSF, solvent-accessible surface area (SASA), hydrogen bonds, and intermolecular contacts. The directory is organized by PDB ID and simulation replica: Feature Data/ └── PDB_ID/ ├── Replica0/ ├── Replica1/ └── Replica2/ 4.Rep_Stru_MD Rep_Stru_MD.tar.gz provides one representative structure for each of the 256 TCR–pHLA complexes. Each structure is extracted from the final frame of Replica0 after equilibration and is provided in PDB format, offering a convenient snapshot for structural inspection and reuse without requiring access to the full trajectories. Rep_Stru_MD/├── 1BD2.pdb├── XXXX.pdb└── ... 5.descriptor.csv descriptor.csv contains structure-level descriptors for the TCR-pHLA complexes. It provides a tabular representation of complex-level structural and biophysical properties for downstream analysis and computational applications. 6.rmsf.csv rmsf.csv contains residue-level root-mean-square fluctuation (RMSF) values derived from the MD trajectories. RMSF is provided as a separate table because it is calculated at the residue level and the number of residues varies among TCR-pHLA complexes. 7. CDR3_pep_contact_freq.csv CDR3_pep_contact_freq.csv provides residue-pair contact frequencies between CDR3β and peptide residues during the MD equilibrium phase (40–50 ns), characterizing the dynamic interfacial interactions of the central recognition region. 8.TCR_IMGT_res_mapping.csv TCR_IMGT_res_mapping.csv provides IMGT49 residue positions, region annotations, and mapping status for residues in the TCR α and β chains, generated by sequence annotation with ANARCI using the IMGT49 numbering scheme. 9.missing_residues_added.csv missing_residues_added.csvrecords missing residues that were identified by sequence alignment and modelled using MODELLER during MD system preparation. Columns: PDB ID, chain ID, chain type (TCR or HLA), residue number, and residue name. This allows users to trace residues added to each structure. 10. Experimental binding data Two files provide complementary experimental binding annotations, matched to DynaTPH by peptide–HLA combination; the corresponding PDB IDs and the number of associated structures are recorded for each combination. TCR_pHLA_affinity.csv— TCR–pHLA binding affinity data from ATLAS for 207 DynaTPH complexes. peptide_HLA_binding.csv— peptide–HLA binding measurements from IEDB for 100 DynaTPH complexes. Both files retain available source and study identifiers, measurement types, and experimental methods and conditions. Multiple measurements are preserved when reported by the source databases. Data Access Zenodo — complete and citable archive of the DynaTPH dataset, including the Dynamic Data. GitHub — lightweight data and directly access, excluding the large Dynamic Data (https://github.com/leifugroup/DynaTPH). Citation Please cite the DynaTPH dataset and the associated research article when using these data. Dataset DOI: 10.5281/zenodo.21971877