This dataset contains all the data used to implement the B-cell epitope predictor tool called HLA-EpiCheck (see https://doi.org/10.1101/2023.12.18.572133).

CONTENTS: - pre-patches: Directory containing the computed pre-patches. A pre-patch corresponds to the set of residues within a given distance from a residue. The patches are generated subsequently by keeping only the solvent-accessible residues. Files are organized by locus and by antigen. A file contains the pre-patches associated to a given residue computed for any frame considered in the trajectory. + patches_resid__size.txt: Each line in the file corresponds to a pre-patch of a given frame. Line format : : Residue numbering is the same as in the PDB files.

• trajectories: Directory contaning the MD data. Files are organized by locus and by antigen.

  • .dcd: 10ns MD trajectory comprising 1000 frames. Water molecules were removed.
  • .psf: Topology file of the .dcd trajectory.
  • .pdb: Starting structure of the MD simulation.

• training_set_size_15.csv: training set used to train HLA-EpiCheck.

• test_set_size_15.csv: test set used to evaluate HLA-EpiCheck.
• table_patch_ID_antigen_residue.csv: Table containing the antigen and central residue associated to each patch.
• model_ERF_radius_15.pkl: ML model of HLA-Epicheck in pickle format. Pickle version 4.0 used.
• descriptors_eplets_non-confirmed.csv: Descriptors of the non-confirmed residue patches.
• preds_non_confirmed_DQ.csv: HLA-EpiCheck predictions on the non-confirmed residue patches of eplets from locus DQ.
• PDB_modeled_structures.txt : List of antigens modeled from a PDB structure with the corresponding PDB entry.

VMD, 1.9.4a57

namd3, 3.0alpha9_gcc-8.3.0

Python, 3.9.13

Scikit-learn, 1.0.2

pickle, 4.0