Version 0.3.2

Fixes bug in CITATION caused by dash (–).

Version 0.3.1

This version contains couple bug fixes for version 0.3.0:

Long titles are now handled with stringr::str_wrap instead of ggtext::element_textbox_simple
Fixed a bug with color legends in scatter plots
Updated outdated examples

Version 0.3.0

Main addition to this release is a new normalization method called Probabilistic Quotient Normalization (PQN). In addition, different visualization related updates have been added.

New functionality:

• Normalize using PQN with pqn_normalization()
• Returning a list of plots instead of saving is supported on save_x_plots() functions

Other changes:

• Plot visuals have been modified slightly and plots offer more customization options
• mz/rt column finding is now more flexible
• Several minor bug fixes

Version 0.2.1

A couple new feautures are added:

0.2.0 introduced possibility to save plots of individual features in EMF format. We then noticed that using one large PDF file can sometimes be useful, so that functionality is added back. In addition, the plots can be saved as PNG, SVG or TIFF images. In addition, text size can be changed. This affects the following functions: save_subject_line_plots() , save_group_boxplots(), save_beeswarm_plots(), save_scatter_plots() and save_group_lineplots()

perform_lm_anova() computes type I ANOVA tables for linear models. More ANOVA models should be added in upcoming releases.

Version 0.2.0

There are two main changes in this release:

• More plot formats are now available, see documentation of individual plotting functions on how to create EMF, SVG, PNG or TIFF images.
• Non-parametric tests are finally available

New functionality:

• save_plot() now supports EMF, SVG, PNG and TIFF images in addition to PDF.
• save_xx_plots() functions now support saving multiple separate files in addition to one multi-page PDF
• perform_mann_whitney(), perform_wilcoxon_signed_rank() and perform_pairwise_non_parametric() implement Mann-Whitney U test and Wilcoxon signed rank tests.

Other changes:

• Column names of results of statistical tests have been changes. All CI columns are now called LCI95 or UCI95. The number can be replaced in cases where the confidence level has been changed via the ... parameter available for some tests
• Confidence level can no longer be set for linear model functions. This option didn't actually work anyway, and it has not been used so it is now removed.

Version 0.1.2

This release contains further improvements to MetaboSet object creation. This fixes the latest bug in #25 and should prevent many similar issues in the future.

Changes:

• construct_metabosets() now makes sure exprs is in correct format.
• MetaboSet validity check contains additional checks to maintain data integrity

Version 0.1.1

This is a very small release, fixing an issue from 0.1.0.

• construct_metabosets() no longer checks m/z or retention time limits

Version 0.1.0

This release adds a couple new features and a bunch of improvements to existing ones.

New features:

• perform_permanova() does permutational analysis of variance using package called PERMANOVA
• perform_pairwise_t_test() has new parameter is_paired which allows paired version of pairwise t-tests
• clean_stats_results() can be used to rename and remove columns from statistics results
• Error messages will be logged to a file after init_log()
• Merging MetaboSet objects will now create dummy (negative incrementing) injection order if injection order differs between modes
• Sample identification column can now be passed to read_from_excel()

Other changes:

• mz_rt_plot() plots can now be filtered by p-values and low-quality features can be dropped from a MetaboSet object
• volcano_plot() plots can be labeled
• cohens_d() is now able to handle multiple comparisons at once
• read_from_excel() will now check that mass-to-charge and retention time values are reasonable, limits can be adjusted
• Column names in statistics results, excluding linear models, follow group + comparison + value style

Bug fixes:

• Simple imputation fails when only one feature satisfies NA limit
• ... arguments will be passed forward in perform_correlation_tests()

Version 0.0.11

The biggest change in this version is that logging now uses a dedicated logging package, futile.logger, which allows for more control over logging levels and better formatted logging messages. Currently, some logging messages are not perfectly formatted for the new way, but this should improve in the future. futile.logger also allows controlling level of logging in the future.

New features

• Improved logging with futile.logger. init_log() no longer overwrites log files.
• fix_MSMS() function creates publication ready name formats for MS/MS spectra

Version 0.0.10

This version has a bunch of improvements that have been on the backlog for a while. Mainly, you can finally write the outputs of statistical analyses to an Excel file and then easily read that Excel file back into R without having to jump through loops to fix awkward column orders etc.

NOTE: this version is developed with R 4.1.2, but should work with R 3.6.3 as well. It is quite likely that notame will only support R 4.x in the future.

Changes:

• construct_metabosets() can now construct a single MetaboSet with all the modes included, instead of forcing the split of analytical modes. Splitting is still the default behavior to ensure backwards compatibility.
• write_to_excel() now handles Datafile columns and the last column of feature data so that the output Excel file is more like the Excel files created by MS-DIAL.
• read_from_excel() recognizes corner columns with 3 letters (previously limited to 2, meaning you would have to calculate the number of a column beyond "ZZ" manually)
• Version 0.0.9 introduced logging messages to read_from_excel(), but some of these messages were also logged when a MetaboSet was modified in any way. This is now fixed and you should not have to bear with extra messages anymore.