← Back to all sparks
S

segregatr

ANALYTICS
Velocity0.0

Segregation Analysis for Variant Interpretation

A segregation-analysis tool that keeps widening which pedigrees it can actually handle.

statistical-geneticspedigree-analysisvariant-classificationpedsuiter-package
Current state
segregatr computes full-likelihood Bayes factors for variant segregation in families, built on pedtools and part of the wider pedsuite ecosystem. Its releases are infrequent but each one lifts a structural restriction: loops, recessive and X-linked models, liability classes, and most recently a proband-free variant of the score. The companion shinyseg app gives the same machinery a clinical front end.
Where it's heading
The through-line is coverage of awkward real-world pedigrees rather than new statistics. Loops were handled for the core score in 0.3.0 and then extended to liability classes in 0.4.0, so the same structural capability is being pushed through the codebase feature by feature. The 2025 release moves in a different direction, relaxing the requirement for a designated proband. Development is slow and steady, roughly annual, and tracks its pedtools dependency closely.
Prediction
Expect the next release to continue relaxing modelling constraints — the pattern of retrofitting each new capability across loops, liability classes and inheritance models is unfinished — rather than expanding beyond segregation scoring.

Recent moves

  1. 11mo ago

    Proband-free variant of the FLB score

    Adds sFLB(), a shared proband-free version of the FLB score. Dropping the designated-proband requirement extends the method to family data where no single individual anchors the ascertainment, which is a genuine widening of the input the package accepts.

    View source ↗
  2. 2y ago

    Liability classes now work in looped pedigrees

    Extends liability class support into looped pedigrees, retrofitting the structural capability added in 0.3.0 to another part of the model. A startdata fix closes a path that had allowed impossible input, and citation links point to shinyseg and the underlying papers.

    View source ↗
  3. 3y ago

    segregatr 0.3.0

    ⚡ SPARK

    The release that made the package broadly applicable rather than restricted: FLB() gained looped pedigrees, recessive and X-linked models, and homozygous carriers in one pass. Later releases have been extending these same capabilities into the rest of the codebase.

    View source ↗