Kinase-phosphatase cycles drive self-sustained oscillations of protein condensates: simulation data, analysis and code

Kinase-phosphatase cycles drive self-sustained oscillations of protein condensates: simulation data, analysis and code Supporting data of > **Zippo, E., Qiao, L., Dormann, D., Speck, T., Schmid, F., Stelzl, L.S. Kinase-phosphatase cycles drive self-sustained> oscillations of protein condensates, bioRxiv (2026), https://doi.org/10.64898/2026.10.27.762139** Coarse-grained molecular dynamics (MD) of the TDP-43 low-complexity domain (LCD) and of full-length TDP-43 with thekinase CK1δ and the phosphatase calcineurin (PP3). (De)phosphorylation reactions are driven by a chemical-potentialdifference Δμ (CALVADOS3 force field, HOOMD-blue v3 with `hoomd3_phosphorylation`). The MD results are compared with acontinuum (phase-field) model. Contents | Folder / file | Content | Size || --------------------------------- | -------------------------------------------------------------------------------------------- | ------------------------------------------ || `simulations/` | inputs and outputs of every simulation, without the full trajectories | 17 GB || `trajectories/` | trimmed trajectories of movies runs (see below) | 19 GB || `movies/` | Supplementary movies from trimmed trajectories | 2.2 GB || `analysis/` | results of the analyses, read by the notebooks to make the figures | 1.7 GB || `plots/` | figures written by the notebooks | 0.2 GB || `code/` | analysis notebooks and the `hpa` package (see `code/README.md`) | 0.06 GB || `phase_field_model/` | data of the phase-field model (the model code is not included) | 0.15 GB || `hoomd3_phosphorylation-3.1_kinase_phosphatase.zip` | MD code used for the simulations, release `3.1_kinase_phosphatase` of https://github.com/ezippo/hoomd3_phosphorylation | 1.2 GB || `LICENSE-DATA.md` | licences | `simulations/`, `analysis/` and `plots/` have one sub-folder per set of runs: | Sub-folder | Simulations || ---------------------------------- | -----------------------------------------------------------------------------------------------------------------|| `droplet_200-tdp43-lcd/` | droplets of 200 LCD chains with N_k CK1δ and N_p PP3: active (`ness`), passive (`dmu0`), CK1δ only, PP3 only || `cycle_single_TDP43/` | one TDP-43 chain with one CK1δ and/or one PP3: enzymatic cycle at different Δμ (200 K) || `condensate_structure/` | droplets and slabs without reactions: LCD or full-length TDP-43 with 5 CK1δ or 5 PP3, CALVADOS3 and HPS-Urry || `fixed_pser_uniform_slabs/` | LCD slabs without reactions, every chain with n pSer (n = 5–11) || `fixed_pser_random_droplets/` | LCD droplets without reactions, on average n pSer per chain, randomly placed || `ostwald_ripening/` | two droplets of total 400 LCD chains with 4 CK1δ and 48 PP3, active and passive || `input_stats/` (in `simulations/`) | force-field parameters, system files and PDB structures | Every run folder of `simulations/` contains the input files (`input*.in`), start and end configurations(`*_start.gsd`, `*_end.gsd`), thermodynamic logs (`*_log.gsd`), the reaction files (`*_contacts.txt`, `*_changes.txt`) when availableand, for `cycle_single_TDP43/ness_ck1d_pp3_masspSer_fr01/` at Δμ_k=-5, Δμ_p=+2.5 , the enzyme–serine distances (`*_dist.txt`) from which the state trajectories are built. Trajectories The full trajectories (`*_dump.gsd`, about 6.7 TB) are kept by the authors and are available on request.`trajectories/` contains trimmed copies of runs relative to Supplementary Movies, with the same folder and file names plus `_trimmed`: all theparticles (positions, images, types, NO velocities), one frame every 5th frame of the original.Additional trajectories are available at the Zenodo record "Additional data: Kinase-phosphatase cycles drive self-sustained oscillations of protein condensates", doi: 10.5281/zenodo.24103647 . Movies - Supplementary Movie 1 : `sim1_ser19_5p-tdp43_ck1d-open_pp3_cycle_trimmed_0_2_24us.mp4`- Supplementary Movie 2 : `sim1_200tdp43_5pp3_recondensation_CAL3_trimmed_centered_3_15us.mp4`- Supplementary Movie 3 : `sim11_400tdp43_4ck1d-open_48-pp3_trimmed_0_3us.mp4`- Supplementary Movie 4 : `sim3_200tdp43_2ck1d_10pp3_oscillations_CAL3_trimmed_centered_0_12us.mp4`- Supplementary Movie 5 : `sim3_200tdp43_3ck1d_15pp3_oscillations_CAL3_trimmed_centered_24_36us.mp4` Re-making the figures Every figure is made by a notebook in `code/` from the files in `analysis/` and `simulations/`; the full trajectoriesare not needed. Install the Python environment (see `code/README.md`), then: ```bashexport HPA_DATA_ROOT=/path/to/this/recordcd /path/to/this/record/codejupyter lab``` and run a notebook from the top. Saved analyses are read from `analysis/`; a step is recomputed only if its file ismissing or `RECOMPUTE = True` is set in the notebook, which needs the full trajectories. | Notebook | Figures ||---|---|| `cluster_size_phospho_notebook` | condensate size and pSer in time, phase diagram of the simulations, cross-correlations, limit cycles || `phosphorates_notebook` | (de)phosphorylation rates, pSer in time with one enzyme type || `droplet_msm_rates` | A/B cycle in and out of the droplet, rates and Δμ, positions of the reactions || `msm_notebook` | enzymatic cycle of a single TDP-43 chain, Δμ contributions || `density_profiles_notebook` | density profiles of droplets and slabs, phase diagram of the fixed-pSer slabs || `contact_maps_notebook` | TDP-43–enzyme contact maps, interaction energy per residue || `ostwald_ripening_notebook` | inputs of the active-droplet theory, Ostwald ripening of two droplets || `phase_field_comparison` | comparison of MD and phase-field model: phase diagram, limit cycles, oscillations | Conventions - **Run names**: `simtype_200-tdp43-lcdN_k-ck1d-openN_p-pp3_fr01/`; `type` = `ness` (active, Δμ_k = −40 kJ/mol, Δμ_p = +8 kJ/mol), `dmu0` (passive, Δμ = 0), `eq` (no reactions); replicas `sim1`, `sim2`, ... inside each folder. Continuations of a run: `_20us_` starts from 20μs, `_40us_` starts from 40μs.- **Units**: time step 0.01 ps (10⁸ steps = 1 µs); lengths in nm; densities in beads per nm³; energies in kJ/mol. Temperature 250 K, 200 K for `cycle_single_TDP43/`.- **pSer**: bead type `SEP`. A/B chains: B if a chain has at least 8 pSer.- **Reaction files** (`*_contacts.txt`, `*_changes.txt`): one line per reaction attempt, changes.txt keeps only accepted changes, `timestep, Ser bead index, acc, distance [nm], dU, enzyme id, x, y, z`; acc = 1 phosphorylation accepted, −1 dephosphorylation accepted, 0 phosphorylation rejected, 2 dephosphorylation rejected. Enzyme ids: kinases first, then phosphatases.- **Analysis files**: `sim*_size_cond_*.dat`: time [steps], chains in the largest cluster, total pSer; `sim*_timelines*_treshold8.json`: A/B and in/out state of every chain in time; `*.npz`: NumPy archives, the arrays are named in the notebook cell that writes them. Licence Data: CC BY 4.0. Code in `code/`: GPL-3.0. `hoomd3_phosphorylation`: see the licence inside the zip file.