Skip to content

KinOpt overview - light mode KinOpt overview - dark mode

KinOpt Documentation

KinOpt estimates cumulative kinase effects on substrate phosphorylation from time-series phosphoproteomics data. This documentation covers the Python modules that are present in this repository:

  • kinopt/
  • processing/
  • config/
  • config_loader.py

DOI

Overview

KinOpt is organized around a kinase-phosphorylation optimization workflow:

  1. Prepare input tables with processing.
  2. Run local or evolutionary optimization with kinopt.local or kinopt.evol.
  3. Review fitted outputs with kinopt.fitanalysis.
  4. Inspect feasibility and optimality diagnostics with kinopt.optimality.
  5. Load runtime settings with config and config_loader.py.

The repository does not include separate project packages for transcription-factor optimization, coupled network simulation, ODE model fitting, or shared utility packages. The docs therefore avoid references to packages that are not present in this checkout.

Package Areas

kinopt

The kinopt package contains the optimization code for estimating kinase effects on phosphorylation trajectories.

  • kinopt.local runs constrained local optimization with JAX/JAXopt-based solver mapping.
  • kinopt.evol runs population-based evolutionary optimization with QDax NSGA2/SPEA2.
  • kinopt.evol enforces alpha/beta bounds and per-group sum-to-one constraints through candidate repair before objective evaluation.
  • kinopt.fitanalysis contains post-fit analysis helpers.
  • kinopt.optimality contains KKT, feasibility, and diagnostic utilities.

processing

The processing package prepares raw data and maps optimization outputs to analysis-friendly tables. It contains:

  • processing.cleanup for input cleanup and transformation.
  • processing.map for result mapping and network table generation.

config

The config package contains runtime helpers:

  • config.cli for command-line mode parsing.
  • config.logconf for logger setup.

config_loader.py

config_loader.py loads config.toml and provides cached access to configuration sections and mode-specific overrides.

Running KinOpt

From the project root:

pixi run kinopt-local
pixi run kinopt-evol
pixi run kinopt-fitanalysis

The equivalent direct Python entry points are:

PYTHONPATH=. python -m kinopt.local
PYTHONPATH=. python -m kinopt.evol
PYTHONPATH=. python -m kinopt.fitanalysis

See the API Reference for import-level documentation.