Architecture¶

Repository shape¶

The current repository snapshot is centered around a single module:

rlto_model.py

That module contains:

data containers,
the main model,
diagnostics utilities,
a demo dataset,
and a script-style demonstration entrypoint.

Main objects¶

`Microenvironment`¶

Encapsulates five environmental stress variables and exposes:

clipping logic,
aggregate stress_index.

`GeneConfig`¶

Defines per-gene parameters and validates biological or numerical constraints.

`ModelResult`¶

A structured result record returned by net_gene_fitness().

`CancerRLTOModel`¶

The primary API surface. It owns:

gene configuration,
environment,
analytical fitness logic,
abundance parameterization,
optimization methods,
pathway summaries,
demo dataset access.

`ModelDiagnostics`¶

A plotting and diagnostics layer built on top of the model.

Internal flow¶

flowchart TD
    A[GeneConfig / dataframe input] --> B[Baseline abundance]
    B --> C[Threshold abundance]
    B --> D[Gamma parameters]
    C --> E[Analytical robustness probability]
    D --> E
    E --> F[Robustness term]
    B --> G[Burden cost]
    B --> H[Toxicity cost]
    F --> I[Net gene fitness]
    G --> I
    H --> I
    I --> J[Evaluation / optimization / intervention]
    J --> K[Tabular outputs and plots]

Dataframe ingestion path¶

from_dataframe() is the practical integration point if upstream data is already tabular.

That makes the current architecture well suited to:

notebook-based exploration,
script pipelines,
synthetic benchmarking,
omics-derived feature tables prepared elsewhere.

Plotting path¶

ModelDiagnostics uses matplotlib and returns figure objects. The example workflow persists them to disk with savefig().

What is not currently present¶

The supplied code does not include:

a package layout,
a command-line interface parser,
configuration files,
logging,
typed schemas for file IO,
tests,
packaging metadata.