Data & Experiments¶

Overview¶

The pipeline is driven by flow cytometry time courses and dose–response experiments for different CasTuner constructs and cell lines.

The key data sources are:

• .fcs files with:
  • non-fluorescent controls (NFC),
  • time-course experiments (repression/derepression),
  • dose–response experiments (dTAG titration).

All input paths are configured in config.yaml.

Experimental conditions¶

1. Degron-controlled repressors¶

Cells express:

• FKBP12^F36V–hHDAC4–dCas9 (CasTuner transcriptional),
• FKBP12^F36V–CasRx (CasTuner post-transcriptional),
• or comparison systems (KRAB-dCas9, KRAB-Split-dCas9).

Each construct includes a tBFP reporter to quantify repressor abundance.

2. Endogenous reporters¶

Endpoints are endogenous genes tagged with fluorophores, e.g.:

• Esrrb-P2A-mCherry in mouse ESCs,
• STAG2–EGFP in HeLa cells,
• Nanog-P2A-mCherry in mouse ESCs.

Readouts:

• mCherry intensity → NANOG/ESRRB levels,
• EGFP intensity → STAG2 levels.

3. Time-course experiments¶

Two core dynamic experiments:

• Repression (upregulation of repressor)
  Cells are shifted from high dTAG (repressor degraded) to lower dTAG.
  Measured over ~6 days:

  • tBFP (repressor) trajectories,
  • mCherry/EGFP (target) trajectories.

• Derepression (degradation of repressor)
  After sustained repression (low dTAG), dTAG is re-added.
  Measured over ~6 days:

  • rapid loss of tBFP,
  • slower recovery of mCherry/EGFP.

These experiments constrain:

• half-times of degron-repressed proteins,
• delays between repressor change and target response.

4. Dose–response experiments¶

Cells are kept long enough at different dTAG doses to reach steady state:

• For each dose:
  • quantify repressor levels (tBFP),
  • quantify target output (mCherry or EGFP),
  • build dose–response curves (normalized BFP vs fold-change in target).

The pipeline fits Hill functions to these curves.

5. Single-cell noise and hierarchical structure¶

The same .fcs data also gives:

• full single-cell distributions of BFP and mCherry/EGFP,
• from which we derive per-condition:
  • mean,
  • variance,
  • CV².

A simple hierarchical model summarises:

• construct-level noise,
• across time points and replicates.

6. Design-space scanning¶

Finally, measured parameters from the above experiments seed a simulated design space:

• we define realistic ranges for K, n, half-times, delays and degradation rate α,
• simulate the ODE model across this space,
• compute summary metrics (dynamic range, t₅₀, t₁₀–₉₀, overshoot),
• and rank designs.

All intermediate tables are written to parameters/ and mirrored into the report.