rel-3-09

• Replaced I-spline rank-based PEP calibration with score-based I-spline calibra
  tion as the new default; use --rank-pep to opt back to the previous rank-based b
  ehaviour. (#403)
• Added I-spline PEP calibration using a trust-region reflective constrained rid
  ge regressor (ISplineTRRRegressor), replacing PAVA post-processing as the defaul
  t smoother; use --pava-pep to revert to PAVA. (#399)
• Removed XML/XSD I/O support, which was incompatible with modern C++ toolchains
  . (#399)
• Fixed a bug where PEP values could exceed 1.0. (#394)
• Fixed a variable scope error in Set::reindexTo. (#402)
• Raise an error when tab characters are found in a non-tab-separated protein fi
  eld. (#380)
• Check that input files exist before attempting decoy prefix detection. (#397)
• Replaced CentOS with AlmaLinux 9 in CI via container build; updated GitHub Act
  ions to Node.js 24 compatible versions and Boost 1.87.0. (#399)

rel-3-08

• Including builder for ghcr.io containers
• Defaulting to Isplines as PEP regressor #391 @percolator
• Functionality for PAVA regression #391 @percolator
• Added some more Unit Tests #391 @percolator
• added libtirpc-dev to ubuntu builder
• Changing build directory structure for Actions
• Actions building with macos-15 runners, including arm64 support #388 @poshul
• updating Boost.Asio interface #388 @poshul
• removing GoogleAnalytics for now,
• Changed paths for triggered builds.
• Adding dependecy to the Eigen library #391
• Removing unit test dependency on pthreads under windows
• Tidied up code #384 @MatthewThe

rel-3-07-01

Fixing the incorrect naming of the rel-3-07 release

rel-3-07

What's Changed

• Implementation of percolator-reset by @percolator in #378
• Added RT to percolator output by @acquayefrank in #374

New Contributors

• @acquayefrank made their first contribution in #374

3.6.5

v3.06.5

• Allow empty pin-file with --no-terminate flag. (#371)
• Skip protein decoy pattern detection if --protein-decoy-pattern is specified. (#356)
• Delay tab file validation and decoy pattern detection to Caller::run().
• Cleaned up and refactored TabFileValidator.
• Deal with DefaultDirection row when merging multiple pin files.
• Better names for portable artifacts.

3.6.4

v3.06.4

• Fix assertion problem with ties in calculating PEPs. (#365)
• Upgrade to macos-12 on GH action builds. (#364, #366)

3.6.3

v3.06.3

• Add check and fallback for <4 bins in BaseSpline PEP estimation. (#362)

3.6.2

v3.06.2

• Take filename into account in TDC (#358)

3.6.1

v3.06.1

• Added checks for number of decoy PSMs<=1 (#337)
• Removing multiple warnings of decoyPrefix not being set (despite being set). (#338)
• #ifdef CRUX macros removed (#340)
• Fixed problem with missing protein IDs. (#342, #343, #329)
• Fix inconsistent behavior of --train-fdr-initial when --trainFDR is zero (#349)
• Extended fix for XercesC ICU dependency to Ubuntu 18.04 (#188)
• Clean up char pointers in tmpFile creation (#346)
• Removed binary reading mode and replace popen by subprocess (#330)
• Better triggered workflow for github actions
• Update GitHub MacOS runner to macos-11 because macos-10.15 is deprecated.

3.6

v3.06

• Added extended unit tests.
• Added code for quicker FDR calculation.
• Updated urls to xerces and google test
• Switched compiler to c++11
• Added basic pepXML support.
• Improved testing in Github actions:
  • Made separate stages for unit and system tests for Ubuntu
  • Introduced a test build for centos
  • Moved unit and system tests to release code.
  • Changed the dependency structure to gtest
• Added command-line argument to switch off reporting to analytics
• Added possibility to include spectrum file name in tab-delimited input, that is carried over to tab-delimited output.
• Added possibility to include retention time in tab-delimited input.
• Discontinued support of the Elude model
• Removed all retention time modeling and all other DOC features