Skip to content

Latest commit

 

History

History
253 lines (140 loc) · 5.13 KB

NEWS.md

File metadata and controls

253 lines (140 loc) · 5.13 KB

MultipleTesting.jl News and Changes

Version 0.6.0

Changes

  • Upgrades julia compat to 1.6 (#127)
  • Upgrade StatsBase compat to 0.34 (#127)
  • Fixes tests where the error type changed in recent julia versions (#127)

Support

  • Supports julia v1.6 and newer

Version 0.5.1

Changes

  • Handle dependencies automatically through CompatHelper
  • Extend support to SpecialFunctions v2
  • Update the readme

Version 0.5.0

Changes

  • Improve the reordering strategy of p-values in adjustment methods. This change saves one sorting step for all adjustment methods that require sorted p-values. As a result, the performance for these methods is significantly improved.
  • Rename p-value combination types, in order to improve consistency with other types.

Support

  • Supports julia v1.4 and newer

Version 0.4.1

Changes

  • Update continuous integration configuration
  • Add support for julia v1.0

Support

  • Supports julia v0.7 and v1.0

Version 0.4.0

Changes

User-facing changes

  • Require julia v0.7 as the minimum version, drop support for julia v0.6
  • Describe package features in the documentation

Support

  • Requires julia 0.7

Version 0.3.1

New Features

  • Build and deploy the package documentation (#94)

Changes

User-facing changes

  • Fix Barber-Candes adjustemt if all p-values are < 0.5 (#93)
  • Update the documentation notebooks (#88)
  • Type the adjust methods (#90)
  • Rename the adjustment field in TwoStep type (#89)

Internal changes

  • Optimise the Hommel adjustment (#91)
  • Enable fast finishing for CI builds (#92)

Version 0.3.0

New Features

  • Convex decreasing density estimator for π0 (#56)
  • Higher criticism scores (HigherCriticismScores) and threshold (HigherCriticismThreshold) estimation (#55)
  • List of publications which describe the implemented statistical methods (#54, #65, #81, #83)

Changes

User-facing changes

  • Fix p-value adjustments with unsorted inputs (#66)
  • Speed up p-value combinations (#75)
  • Define return types for p-value combinations (#76)
  • Unify methods for weighted Stouffer combination (#84)
  • Update documentation notebooks (#67)
  • Drop support for julia 0.5, require julia 0.6 as the minimum version (#52, #64, #58, #82)

Internal changes

  • Simplify the step functions for p-value adjustments (#77)
  • Simplify the method structure (#74)
  • Switch to new julia syntax, with preliminary support for julia 0.7 (#71, #78)
  • Update tests and references for p-value combinations (#70)
  • Use clamp for better bounding of estimates (#85, #82)
  • Support the new Distributions interface (#52)
  • Drop the Compat dependency (#53)
  • Link comic logo in the readme file as an external resource (#69)

Removed

  • Drop qvalues in favour of BH-adjusted p-values (#73).

Support

  • Requires julia 0.6
  • Preliminary support for julia 0.7 nightly builds

Version 0.2.3

Changes

  • Switch to new Weights type in 'StatsBase' (>= v0.15.0)

Version 0.2.2

Changes

  • Makes the PValues type immutable and persistent
  • Finalises support of julia 0.6 with testing

Support

  • Requires julia 0.5 or 0.6

Version 0.2.1

Changes

  • Adds compatibility with julia 0.6 development versions, including automated tests
  • Fixes Censored BUM π₀ edge case when all p-values are equal to 1
  • Updates test cases and reference values for qValues
  • Introduces Compat as a new dependency
  • Separates test coverage reports for different julia versions

Support

  • Requires julia 0.5 or 0.6

Version 0.2.0

New Features

  • Combination of p-values through the combine interface:
    • Fisher
    • Stouffer
    • Logit
    • Wilkinson
    • Simes
    • Tippett
    • Minimum of adjusted p-values
  • Analysis notebook for p-value combinations
  • PValues type: Representation of vectors with p-values
  • Adjustments of p-values support specifying the total number of tests
  • Barber-Candès p-value adjustment

Changes

  • Optimised Benjamini-Liu p-value adjustment and test cases
  • Estimator types are now immutables
  • Harmonised type names
  • Restructed readme
  • Simplified method signatures
  • Various performance improvements

Removed

  • Export of lower-level functions for p-value adjustment

Support

  • Requires julia 0.5

Version 0.1.0

New Features

  • New π₀ estimators: BUM, Censored BUM, Flat Grenander
  • New p-value adjustment methods: Forward Stop, Benjamini-Liu
  • BUM model for simulating p-value distributions
  • Fast isotonic regression
  • Reference list of related software packages in documentation

Changes

  • Grenander ECDF estimator
  • Immutable method types
  • Default α value for Two Step constructor
  • Tests use new julia 0.5 base testing framework
  • Documentation moved to 'docs' directory

Removed

  • Test coverage reporting through coveralls

Support

  • Requires julia 0.5

Version 0.0.2

New features

  • New π₀ estimators: Two Step, Right Boundary
  • New p-value adjustment methods: Benjamini-Hochberg Adaptive, Oracle, Sidak
  • Package precompilation

Changes / Changed

  • Redefined Oracle Benjamini-Hochberg as Adaptive Benjamini-Hochberg procedure

Support

  • Requires julia 0.4