Skip to content

v1.0.2
Compare
Choose a tag to compare
@nolan-h-hamilton nolan-h-hamilton released this 12 Sep 04:11
· 5 commits to main since this release
v1.0.2
7293b57

v1.0.2

  • Several substantial changes to the CLI and modules with according documentation updates
  • Additional features for scaling and filtering samples' count data prior to scoring
  • Multiple options to measure central tendency and dispersion during scoring:
    • Central tendency: median(still default), trimmed mean via scipy.stats.tmean, general np.quantile, traditional mean, etc.
    • Dispersion: median absolute deviation (still default), trimmed standard deviation via scipy.stats.tstd, iqr, etc.
  • Post-scoring parametric-sigmoid transformation (rocco.parsig(), --use_parsig)
    • Can be useful to promote integrality of solutions from the relaxation directly
  • Split modules for obtaining read count tracks from BAM files (readtracks.py)
  • ROCCO now relies exclusively on OR-Tools to solve the relaxed optimization problem
    • glop -- simplex method, ensures vertex solutions
    • pdlp (default) -- first-order PDHG method scalable to massive problems. Minimal accuracy degradation in the context of ROCCO.
Assets 2
Loading