Bug fix

• Fix qntmap() raising error on the Mapping conditions pane of the Input page, “Can’t create column Values: Can’t combine ..1 and ..2 ” (#34).

• Fix cluster_xmap() raising error when checking unit of pixels (#36)

Internal changes

• Use GitHub Actions for CI (#35)
• Updated tests and DESCRIPTION to pass CI (#37)

Bug fix

• Fix beta being wrong in the Params for Quantify page.

Planning

• Add remove_outliers option to quantify and find_centers

Biggest NEWS

• Graphical user interface is ready! Run qntmap() to enjoy it.

Breaking changes

• qntmap() default to use shiny web app. Specify shiny = FALSE to use legacy console-based intearactive mode. Note that the legacy one is no more maintained.
• read_qnt()
  • omits the renew option
  • uses the saving option only to save phase list, no more to cache data
• read_xmap()
  • omits the saving and the renew option
  • returns a data frame with columns x and y indicating coordinates of pixels instead of a list
• cluster_xmap()
  • returns a data frame with the center attribute unlike older one returned a list
  • omits group_cluster option. Instead, apply group_subclusters() to the result to integrate subclusters.
  • saves the result
    • as is in the formats of RDS, png (dot-by-dot phase map), and svg (phase map with legend)
    • converted by group_subclusters in the format of only png and svg
• group_subclusters()
  • is defined to defunct group_cluster()
  • returns a data frame in the same format as cluster_xmap() but lacks the center attribute.
• pick() changed its first argument to .df from .data.
• The magma color scale become the default one to plot of mapping data instead of the viridis.
• Use message rather than cat so that they can be suppressed by users.

Major changes and new features

• autoplot() methods are implemented to follow the best practice of ggplot2. From now on, plot() methods are wrappers of autoplot.
• find_outlier() is implemented to detect outliers cause mainly by mapping fine grained phases.
• find_outlier() is implemented to detect unexpectedly high or low mapping intensities compared to spot intenstieis, which typically occurs due to outliers are analysing multi-phase pixels
• find_centers()
  • has more robust approach to find centroids powered by find_outlier()
  • has additional arguments
    • phase and element: Specifying them controls which phases and elements are used to find outliers. They suppot tidyeval. For example element = c(Si, Ti, Al) indicates the 3 elements are used, and element = c(-Si, -Ti, -Al) indicates the 3 elements are NOT used.
    • epma: If tidy_epma is already performed, then specify the result here to save time.
  • deprecates fine_phase which is generalized by phase.
• tidy_epma has new arguments subcluster and suffix.
• hmean() and vmean() supports the objects with class qntmap, qm_xmap, and qm_cluster.mPreviously they only supported the qntmap class object.
• All layers for ggplot2 are no longer pre-compiled at installation stage. This is to avoid possible conflicts on updates on ggplot2 after installing qntmap.
• quantify() get the params attribute showing the parameters, alpha, beta, and gamma.
• autoplot.qm_epma() is defined for the result of tidy_epma()

Internal changes

• tidy_epma() returns qm_epma class object.
• gghist.numeric()
  • draws border to be legible regardless of background colors.
  • supports a histogram with a single column.
• correct_deadtime() supports re-calculation of dead time.
• prioritize() accepts unnamed vector.
• save4qm() methods, summary() methods, pick(), mean(), vmean(), and hmean() are re-written to support new data structures.
• correct_deadtime() is factored out from read_xmap(), but not exported currently.
• Themes for plot supports modifying base_size, so that the appearance can be optimized in Shiny App.
• Full support of tidyeval
• Tidyverse and r-lib related functions are imported in qntmap-package.R to avoid importing the same funcions in multiple files.
• Tests run faster by exporting duplicated codes to setup.R.

New features

• cluster_xmap() passes ... to PoiClaClu::Classify(), enabling
  • training with selected data
  • testing with selected data
  • fine tuning on parameters
  • and so on.
• Added vmean() and hmean() to calculate profile of mean values of mapping data. vmean() stands for vertical mean and hmean() for horizontal mean.

Major changes

• Drastically improved performance of plot()
• Clustering results will be saved as an RDS file, a dot-by-dot png image without legend, and a ggplot2 plot with legend saved as svg.
  • The size of ggplot2 plot is set to A4 (297 mm * 420 mm), either horizontal or vertical which fits best to the plot.
  • The legend used to be saved as a pie chart.

Breaking changes

• quantify() no more calculates standard errors by default. This is to save time and memory usage. Set se = TRUE to calculate the errors.
• Format of “params.csv” created by quantify() is changed.
  • Column “elem” is renamed to “oxide”
  • Column “elint” is renamed to “element”
  • Columns "*_se" are removed.

New features

• quantify() and tidy_epma() supports maps and spots analyzing inequal elements.

  Example	Mapped elements	Spot elements
  A	Si, Ti, Mg	Si, Mg
  B	Si, Mg	Si, Ti, Mg
  C	Si, Ti, Mg	Si, Mg, Ca

Bug fix

• Calculation of standard errors for compositions of pixels were wrong.

Major changes

• quantify() saves parameters alpha, beta, gamma, and their standard errors when saving = TRUE.
• quantify() loads parameters from a csv file by giving its path to fix parameter.
• read_qnt() and read_xmap() supports to retrieve analytical conditions from manually prepared csv files. This feature is implemented to avoid failures in reading data caused by difference in formats output by different EPMA instruments.
• Hover on plot() of mapping data shows coordinates and a chemical composition of a selected element.

Major changes

• quantify() supports fixing compositions of certain components of phases.
• Defunct deprecated functions:
  • qltmap_cls_centers
  • qltmap_load
  • qnt_load
  • qntmap_cls_pois
  • qntmap_quantify
• Support for searching online documents: https://qntmap.atusy.net

Minor changes

• Refactored to
  • simplify codes
  • separate monster functions into small ones
• Introduced continuous integrations:
  • Travis CI, AppVeyor, and Codecov
  • As Travis CI passes building on macOS, #23 is closed.

Decisions

• qntmap() will be maintained without new features to focus on developing web user interface with shiny package.

Major changes

• plot()
  • Interactive plot is a combination of shiny, ggplot2, and more. No more using plotly.
    • Color control with histogram
    • Hover information of pixels
    • Zooming
    • Moving
    • Summarizing data in a table below heatmap
      • Double click to keep pixel data
      • Box select to keep mean values of data
  • Grayscale is supported in addition to viridis. #28
  • Clustering result is supported. #28
• segment()
  • Only allows PNG format as an input. No more JPG nor BMP allowed.
  • Independent from imager package. #16

Major changes

Minor changes

• read_cnd, read_map_pos, and read_map_beam are no more in use but are remained for awhile. Note they are not exported functions.

New features

• Visualize X-ray maps and quantitative maps using a plot function.
  • Parameter shiny = TRUE gives rich interactive features although still under development.
• Functions to add phases manually in case user finds some phases not quantified after EPMA analysis. See https://qntmap.atusy.net/articles/add_phase.html for more detail.
• mean to calculate mean value of quantitative maps.
  • Parameter index enables index or mask based calculation of mean. See https://qntmap.atusy.net/articles/qntmap.html#summary-based-on-mask-images for more detail.

Breaking changes

• Renamed functions to follow verb_hoge style. This leads some functions to be deprecated or defunct. See “Summary of defunct, deprecated, and unexported functions”.
• Unexported cipois and flag0.
• Renamed arguments to be shorter and easy-remembering.
• Reduced arguments which are reused again and again by utiliing attributes. For example,
  • Path to the directory containing X-ray map data is specified by dir_map or wd in use of epma_tidy, qltmap_cls_centers, qltmap_cls_pois, qltmap_load, and so on. However, now is only specified in read_xmap.
• The above change cause read_qnt to save its result (qnt.RDS) and semi-product (phase_list0.csv) in the directory containing .qnt files unlike qnt_load saved them under the working directory.

Major changes

• Monster functions are separated into smaller functions.
  • A wrapper part of PoiClaClu::Classify in cluster_xmap (formerly qltmap_cls_pois) is isolated as cluster
  • Internal parameters for quantify are calculated using find_AB, find_AG, and find_B
  • Codes for saving result of functions are isolated as S3 generics of save4qm.
• Condition files of EPMA analysis are read by S3 generics to support various types of file formats.
• find_centers (former qltmap_cls_centers) become more robust against multi-phase pixels by - finding pixels analyzing multi-phase pixels by comaring with corresponding quantified points. - predicting X-ray map counts of a target phase in multi-phase pixels based on regression analysis.
• cluster_xmap does not group_clusters by default.

Minor changes

• Performance improvements by simplifying codes, utilizing matrixStats package, and so on.
• quantify supports fixing internal parameters (A and B).

Summary of defunct, deprecated, and unexported functions