Dataset Viewer

The dataset viewer is not available because its heuristics could not detect any supported data files. You can try uploading some data files, or configuring the data files location manually.

BENDER-MPIPI

Companion dataset to BENDER. Contains coarse-grained molecular dynamics trajectory data from MPIPI (Multi-scale Platform for Intrinsically disordered Protein Interactions) simulations of intrinsically disordered proteins (IDPs).

Contents

Trajectories are organized by taxon. Each zip file contains subdirectories named by UniProt ID, with the following files per protein:

  • traj.xtc — trajectory file
  • log.lammps — LAMMPS simulation log
  • Rg.out — radius of gyration output
  • run.in — LAMMPS input script
  • run.sh — run script
  • seq.dat — sequence data

Sequences per Taxon

Taxon Sequences
Bacteria 318
Fungi 179
Mammals 164
Plants 163
Protists 159
Viruses 163
Total 1,146

Simulation Details

All simulations use the Mpipi-GG (v23-r1) coarse-grained force field (Lotthammer et al., Nat Methods 2024), run with LAMMPS. The protocol mirrors the CALVADOS IDRome adaptive scheme:

Force Field & Representation

  • One-bead-per-residue coarse-grained model (C-alpha representation)
  • Bond potential: Harmonic, k = 9.6 kcal/mol/A^2, r0 = 3.81 A
  • Non-bonded interactions: Wang-Frenkel (WF) potential with cutoff 25.0 A
  • Electrostatics: Coulomb/Debye screening at 150 mM implicit salt (kappa = 0.126 A^-1)
  • Charged residues: K, R (+0.75e); D, E (-0.75e)
  • Dielectric constant: 80.0

Simulation Protocol

  • Integrator: NVE + Langevin thermostat (equivalent to NVT)
  • Temperature: 300 K, damping = 5000 fs
  • Timestep: 20 fs
  • Boundary conditions: Periodic (PBC)
  • Box size: 500 A for sequences <= 300 residues; (N/5 + 20) x 10 A for longer sequences
  • Initial configuration: Self-avoiding random walk, followed by energy minimization

Adaptive Simulation Length

  • Sequences <= 150 residues: 200 ns total
  • Sequences > 150 residues: 200 x (N/150)^2 ns (quadratic scaling)
  • Equilibration: First 50% of trajectory discarded
  • Production frames: 10,000 (<=300 ns), 15,000 (<=600 ns), or 20,000 (>600 ns)

Output

  • traj.xtc — XTC trajectory (production frames only, post-equilibration)
  • Rg.out — Radius of gyration time series
  • log.lammps — Full simulation log with thermodynamic properties
  • seq.dat — LAMMPS data file with sequence, topology, and initial coordinates
  • run.in — Complete LAMMPS input script (fully reproducible)

Citation

@misc{sabian98_2026,
    author       = { Taseef Rahman },
    title        = { BENDER-MPIPI (Revision 57dd8bf) },
    year         = 2026,
    url          = { https://huggingface.co/datasets/taseef/BENDER-MPIPI },
    doi          = { 10.57967/hf/10438 },
    publisher    = { Hugging Face }
}

Per-sequence data (added 2026-10-25)

Two tables in data/ give, for the 1,146 proteins in the zips, the ensemble properties computed with the BENDER estimators from the production half of each trajectory.

  • data/mpipi_results_v2.csv has one row per protein, with the apparent scaling exponent (nu_mpipi, its fit R², block standard error and the values from each half of the production run), the radius of gyration and end-to-end distance in nm with their errors, asphericity, residue-pair contact frequencies (for example Arg and Lys with aromatic and acidic residues), sequence descriptors (SCD, NCPR, FCR, Arg, Lys and aromatic fractions) and contact-network properties. status is missing for 4 proteins whose analysis did not complete.
  • data/mpipi_graph_features_v2.csv has the contact-network properties of 1,142 proteins (fragmentation index, global efficiency, average clustering, transitivity and degree assortativity, each with its standard deviation over frames).

The kingdom column carries the taxonomy check of 2026-09-27, so 37 proteins stored in Protists.zip are labelled there as nematodes or other invertebrates. The 957 proteins compared with CALVADOS-2 in the BENDER paper are those with in_paper_set true in the six taxa Bacteria, Fungi, Mammals, Plants, Protists and Viruses.

Mpipi-Recharged per-sequence data (added 2026-10-26)

data/mpipi_recharged_300K_features.csv gives, for all 10,282 proteins of the BENDER release, the ensemble properties of the Mpipi-Recharged simulations at 300 K and 150 mM, computed with the same estimators as for CALVADOS-2 from the production half of each trajectory. Each row holds the taxon (kingdom), the apparent scaling exponent (nu, with its fit R², block standard error and the nu_r2_pass flag for fits with R² of at least 0.99), the radius of gyration, end-to-end distance and asphericity with their errors, contact-network properties, π–π and cation–π contact frequencies, residue-pair contact frequencies and the agreement with AlphaFold2 predicted aligned errors (pae_r, available for 9,983 proteins). Three sequences contain an unknown residue (X), which was simulated as Gly as in CALVADOS-2. The Mpipi-Recharged trajectories themselves are not yet part of this dataset.

Downloads last month
58