HD Phase Separation Analysis

Overview

This repository contains the computational reproducibility code for:

mHTT Low-Complexity Domain Phase Separation Sequesters Transcriptional Regulators: Computational Evidence for Condensate-Targeted Therapy in Huntington's Disease — a Flory-Huggins polymer-mixing analysis of mutant Huntingtin (mHTT) exon 1 liquid-liquid phase separation (LLPS), calibrated to published experimental data, predicting polyQ-length-dependent transcription factor sequestration and prioritising condensate-dissolving compounds for biochemical screening.

The full preprint (PDF + LaTeX source) is included under docs/.

Repository Structure

.
├── README.md
├── LICENSE                                  # MIT license
├── requirements.txt
├── analysis/
│   └── hd_phase_separation/
│       └── hd_phase_separation_calvados.py  # Flory-Huggins phase diagram, R_g, TF partitioning
└── docs/
    └── hd_phase_separation_paper.pdf / .tex

Installation & Usage

pip install -r requirements.txt

python analysis/hd_phase_separation/hd_phase_separation_calvados.py \
  --discovery_id standalone --polyq 23 36 46 72

The script writes figures, CSV outputs, and a run manifest to an output directory (see --help for options).

Key Findings

• Q46 mHTT crosses the LLPS phase boundary at ~3.5 µM (Flory-Huggins model calibrated to Peskett et al. 2018); Q23 wildtype does not phase-separate at physiological concentrations.
• SP1, CBP/p300, and TFIID/TAF4 — all with published mHTT co-localisation — partition strongly into the condensate phase (coefficients 3.6–5.8), consistent with the 38–52% target-gene suppression observed in HD striatum.
• JQ1/OTX015, mitoxantrone, and rifampicin are prioritised as condensate-dissolving candidates for biochemical screening; this is a therapeutic axis distinct from the field's dominant HTT-lowering strategies (ASOs, splice modifiers, PROTACs).

Citation

If using this code, please cite the preprint in docs/.

License

MIT License — see LICENSE file for details.

Contact & Collaboration

Author: John Goodman Email: john.goodman@solver.press Organization: AegisMind Research

We welcome collaboration with experimental validation groups on biochemical FRAP/co-IP screening for the condensate-dissolving compounds identified here.