kombinekm 2.0.0

KoMbine is a software package for propagating statistical and systematic uncertainties in a biomedical analysis. KoMbine provides Kaplan-Meier curve analysis functionality, while ROC Picker provides similar functionality for ROC curves.

ROC Picker and KoMbine

Welcome!

ROC Picker is a software package for propagating statistical and systematic uncertainties in a biomedical analysis using ROC curves.

KoMbine provides Kaplan-Meier curve analysis functionality with uncertainty propagation using likelihood-based methods.

Full documentation

For detailed information and examples, please see the documentation. The source is in the docs/ folder, and you can download the output (latest version from the main branch, compiled by Github Actions) from this link for KoMbine and from this link for ROC Picker.

Quick start

To install ROC Picker and KoMbine, clone the repository, enter its folder, and do

pip install .

Here is a simple ROC Picker example:

from roc_picker.discrete import DiscreteROC
responders = [1, 1, 2, 3, 9, 10]
nonresponders = [2, 3, 3, 4, 6, 8, 9, 10, 10, 10, 10, 11, 12, 13]
DiscreteROC(
  responders=responders,
  nonresponders=nonresponders,
).make_plots(
  npoints=100,
  yupperlim=20,
  #if you want to save the output plots
  filenames=("roc.pdf", "auc_scan.pdf", "roc_errors.pdf"),
  #if you're running in a jupyter notebook or similar, and want to see the plots
  show=True,
)

Here is a simple KoMbine example:

from kombine.datacard import Datacard
import pathlib

# Load a datacard with Kaplan-Meier data
datacardfile = pathlib.Path("test/kombine/datacards/simple_examples/simple_km_few_deaths.txt")
datacard = Datacard.parse_datacard(datacardfile)

# Create Kaplan-Meier likelihood curves for low and high parameter groups
kml_low = datacard.km_likelihood(parameter_min=-float('inf'), parameter_max=0.45)
kml_high = datacard.km_likelihood(parameter_min=0.45, parameter_max=float('inf'))

# Plot the results
kml_low.plot()
kml_high.plot()