Quick Start¶

This guide walks through the complete pipeline from raw .RData inputs to PDE model outputs, using the five ovarian cancer patients included in the repository.

1. Extract patient data¶

Convert the raw liquidCNA .RData files into per-patient CSV archives:

tumorfits extract-data \
    --data-root data/ \
    --out-dir   data/patient_data

This walks data/CNA_tables/ and data/liquidCNA_results/, reads each .RData file, and writes every data.frame object as a CSV to data/patient_data/<patient_id>/.

2. Fit the ODE model¶

Fit the well-mixed ODE model to all patients with Accept_estimate = yes or maybe:

tumorfits ode \
    --data       data/liquidCNA_results/Subclonal_ratio_estimates.extended.txt \
    --flag       yes,maybe \
    --sample_list data/OV_patientDNA_sampleList.txt \
    --use_ca125_updated \
    --drop_failed \
    --out_points ode_gof_points.csv \
    --diag_dir   results/ode_diag \
    --n_starts   8 \
    --maxiter    1200

The output ode_gof_points.csv is a long-table CSV with one row per parameter per patient. It is the required input for all subsequent steps.

3. Generate PDE heatmaps¶

Visualise the 1-D reaction–diffusion dynamics without re-fitting:

tumorfits heatmap \
    --data       data/liquidCNA_results/Subclonal_ratio_estimates.extended.txt \
    --ode_points ode_gof_points.csv \
    --patient    UP0018 \
    --out_dir    results/heatmaps

4. Run the PDE model¶

Run the full 1-D PDE model (optionally fitting diffusion coefficients):

tumorfits pde \
    --data        data/liquidCNA_results/Subclonal_ratio_estimates.extended.txt \
    --ode_points  ode_gof_points.csv \
    --patient     ALL \
    --out_dir     results/pde \
    --n_cells     200 \
    --DS          0.01 \
    --DR          0.01

5. Generate mesh visualisations¶

Run the 2-D FEniCS simulation and produce PyVista PNG visualisations:

tumorfits mesh-view \
    --data       data/liquidCNA_results/Subclonal_ratio_estimates.extended.txt \
    --ode-points ode_gof_points.csv \
    --out-dir    results/mesh_view \
    --patient    ALL

Run the full pipeline with Snakemake¶

All steps above are wired into a Snakemake workflow driven by config.yaml. Edit config.yaml to customise paths and parameters, then run:

snakemake --cores all --configfile config.yaml

Dry-run first:

snakemake -n --configfile config.yaml