# ChangeLog

All notable changes to this project will be documented in this file.

The format is based on [Keep a Changelog](https://keepachangelog.com/en/1.0.0/),
and this project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0.html).

## [1.4.1] - 2022-11-15
### Update
- Changed to support the new version of spatstat

## [1.4] - 2021-10-17
### Fixed
- Fixed a ggplot2 warning when plotting msImage objects.

### Added
- function to generate RGB image of first principal components as msImage
- ***NEW*** msiDataset contains total-ion-count and num. detected ions image that can be used
  to align the calculated reference images
- ***NEW*** now the continuous and binary reference images are in two separate functions
- parallel calculation using SNOW

## [1.3.3] - 2021-05-17
### Fixed
- Removed calls to closeAllConnections()

## [1.3.2] - 2021-03-22
### Fixed
- Fixed dependencies from new spatstat packages.

## [1.3.1] - 2021-01-29
### Fixed
- Now K-means clusters are calculated from the first 10 principal components
  scores. This fixes the incompatibility between stats.kmeans and long format
  vectors.

## [1.3] - 2020-03-21
### Fixed
- Removed dependency from SDMTools, moved to imager.

## [1.2] - 2019-7-19
### Added
- New fields in the msiDataset object save the applied normalization and variance
  stabilizing transformation.
- New normalization methods:
  - Trimmed mean M-values (TMM). Pixel-wise scaling normalization introduced for
    RNA-seq in edgeR.
  - UpperQuartile. Pixel-wise scaling normalization (introduced in edgeR).
- New variance stabilizing transformation (VST):
  - Centered log-transformation (CLR). When TIC scaling is used for normalizing the
    peak intensities, a CLR transformation must be applied to take into account of
    the compositional nature of the data.
- Custom offset for zero intensity peaks in normalization and VST methods.

### Various improvements.

## [1.1.1] - 2018-10-25
### Fixed
- Fixed bug in the global reference filter that won't allow the calculation of
  the similarity measures when constant columns are present in the intensity matrix.

## [1.1] - 2018-10-19
### Added
- New ROI detection using linear SVM. The user must select representative regions
  of the off-sample and sample-related areas. Then, a linear SVM performs the
  segmentation of the data, generating the ROI.
- New ROI detection using k-means with a larger number of clusters than 2. This
  allows a finer detection of the sample-related region.
- Now the covariate image for the Kolmogorov-Smirnov test can be passed as argument
  of the function 'CSRPeaksFilter' (see doc).
- New functions 'addBorderImage' and 'remBorderImage' to add or remove a border
  of N pixels from an MS image.
  
### Fixed
- Fixed the SSIM function (it does not require scaling)
- Fixed typos in the help

## [1.0.4.1] - 2018-10-08
### Removed
- Removed dependency from 'autothresholdr' package. Now Otsu is performed using
the function threshold(x, 'auto') from 'imager'.

## [1.0.4] - 2018-10-06
### Fixed
- Fixed a bug in the function .match.mz.array.
- Improved the comments in the function .match.mz.array.