NEWS.md

metan (development version)

Minor improvements

• Fix bug in get_model_data() calling objects of class mgidi with what = "PCA".
• Improve output of gmd() for gge objects.
• New option projection in gmd() for gge objects to get the projection of each genotype in the AEC coordinates.

metan 1.13.0

New functions

• progress() and run_progress() for text progress bar in the terminal.
• rbind_fill_id() To implement the common pattern of do.call(rbind, dfs) with data frame identifier and filling of missing values.
• hmgv(), rpgv(), hmrpgv(), blup_indexes() to compute stability indexes based on a mixed-effect model.
• mps() and mtmps() for uni- and multivariate-based mean performance and stability in multi-environment trials.

Minor improvements

• ge_reg() now returns hypotesis testing for slope and deviations from the regression. Thanks to @LeonardoBehring and @MichelSouza for the suggestion.
• Resende_indexes() now remove NAs before computing harmonic and arithmetic means.
• Improved outputs in plot_scores that now has a highlight argument to highlight genotypes or environments by hand. Thanks to Ibrahim Elbasyoni for his suggestions.
• Licecycle badges added to the functions' documentation.
• Fix bug in clustering() when using with by argument and defacult nclust argument.
• get_model_data() now extract BLUEs from objects computed with gamem() and gamem_met(). Thanks to @MdFarhad for suggesting me this improvement.
• g_simula() and ge_simula() now have a res_eff to control the residual effect.
• mgidi() now have an optional weights argument to assign different weights for each trait in the selection process. Thanks to @MichelSouza for his suggestion.

metan 1.12.0

New functions

• get_levels_comb() to get the combination of the levels of a factor.
• g_simula() to simulate replicated genotype data.
• add_row_id() to add a column as the row id.
• remove_rownames(), column_to_rownames() and rownames_to_column() to deal with rownames.

Minor improvements

• New argument sel.var() in corr_ci() to filter correlations with a selected variable
• New arguments fill and position.fill in plot_ci() to fill correlations by levels of a factor variable.
• Remove deprecated arguments in arrange_ggplot() and gge().
• New argument theme in arrange_ggplot() to control the theme of the plot.
• Include by argument in gafem().
• mgid() now understands models of class gafem_grouped.
• Fix bug in get_levels() to get the levels even if the variable is not a factor.

metan 1.11.0

New functions

• set_union(), set_difference() and set_intersect() for set operations with many sets.
• venn_plot() to produce Venn diagrams.

Minor improvements

• gge() now have a by argument and understand data passed from group_by.
• New arguments col.stroke and size.stroke in plot.gge()
• gtb and gytb now produces biplots with lines for genotype's vectors in type = 1.
• get_model_data() now understand objects of class fai_blup and sh.

metan 1.10.0

New functions

• get_dist() to get distance matrices from objects of class clustering.
• get_corvars() to get normal, multivariate correlated variables.
• get_covmat() to obtain covariance matrix based on variances and correlation values.
• as_numeric(), as_integer(), as_logical(), as_character(), and as_factor() to coerce variables to specific formats quickly.
• n_valid(), n_missing(), and n_unique() to count valid, missing, and unique values, respectively.
• tidy_colnames() to clean up column names. It is a shortcut to tidy_strings() applied to the column names of a data frame.
• env_stratification() to perform environment stratification using factor analysis.

Minor improvements

• as_*() now handles vectors.

• plot.corr_coef() now shows both stars or p-values for reporting the significance of correlations.

• gamem(), gamem_met(), and waasb() now have a by argument and understand data passed from group_by.

• mtsi() and mgidi() now returns the ranks for the contribution of each factor and understand models fitted with gamem() and waasb() using the by argument.

• plot.mtsi() and plot.mgidi() now returns a radar plot by default when using type = "contribution".

• get_model_data() now returns the genotypic and phenotypic correlation matrices from objects of class waasb and gamem.

• replace_string(), replace_number(), extract_string(), and extract_number() now accepts tidy evaluation in the new ... argument.

metan 1.9.0

• New functions add_prefix() and add_suffix() to add prefixes and suffixes to variable names, respectively.
• New function select_pred() to selects a best subset of predictor variables.
• New function acv() to compute the adjusted coefficient of variation to account for the systematic dependence of $\sigma^2$ from $\mu$.
• New function ge_acv() to compute yield stability index based on the adjusted coefficient of variation.
• New function ge_polar() to compute yield stability index based on Power Law Residuals (POLAR) statistics.
• New function mantel_test() to performs a Mantel test between two matrices.
• New arguments prefix and suffix in concatenate() to add prefixes and suffixes to concatenated values, respectively.
• List packages providing the Rd macros in 'Imports' instead of 'Suggests' as suggested by the CRAN team.

metan 1.8.1

• Use \doi{} markup in Rd files.

metan 1.8.0

• New function gytb() to generate the Genotype by yield*trait biplot.
• New functions row_col_mean() and row_col_sum()to add a row with the mean/sum of each variable and a column with the mean/sum for each row of a matrix or data frame.
• New functions has_zero(), remove_cols_zero(), remove_rows_zero(), select_cols_zero(), select_rows_zero(), and replace_zero() to deal with 0s in a data frame.
• Fix bug of inconsistent color legend when plotting objects of class gge.
• Include class gge and can_corr in get_model_data().
• New argument position in plot.gamem() and plot.mtsi() to control the position adjustment of the bar plot.
• New argument col.by in corr_plot() to map the color of the points by a categorical variable.
• New argument use_data in functions mgidi(), fai_blup(), and Smith_Hazel() to control which data is used (BLUPs or phenotypic means) to compute the index.
• inspect() now generate a warning if zero values are observed in data.

metan 1.7.0

• New functions clip_read() and clip_write() to read from the clipboard and write to the clipboard, respectively.
• New function sum_by() to compute the sum by levels of factors.
• Update wsmp.R (#7). Thank you @BartoszKozak for your contribution.
• mgidi() now allows using a fixed-effect model fitted with gafem() as input data.
• round_cols() now can be used to round whole matrices.

metan 1.6.1

• plot.mgidi() can now plot the contribution for all genotypes.
• plot_bars() and plot_factbars() now shows the values with values = TRUE
• Update the functions by using the new dplyr::across()
• Update citation field by including number and version of the published paper.

metan 1.6.0

New functions

• Smith_Hazel() and print.sh() and plot.sh() for computing the Smith and Hazel selection index.
• coincidence_index() for computing the coincidence index.

Minor improvements

• get_model_data() now extracts the genotypic and phenotypic variance-covariance matrix from objects of class gamem and waasb.
• fai_blup() now returns the total genetic gain and the list with the ideotypes' construction.
• mgidi() now computes the genetic gain when a mixed-model is used as input data.
• The S3 method plot() for objects of class mgidi has a new argument type = "contribution" to plot the contribution of each factor in the MGIDI index.
• plot_scores() now can produce a biplot showing other axes besides PC1 and PC2. To change the default IPCA in each axis use the new arguments first and second.

metan 1.5.1

Minor changes

• plot_bars() and plot_factbars() now align vertically the labels to the error bars.
• fai_blup() now returns the eigenvalues and explained variance for each axis and variables into columns instead row names.
• Fixes the error with donttest{} examples. Now, the correct data set is used in the example of fai_blup()

metan 1.5.0

New functions

• select_rows_na() and select_cols_na() to select rows or columns with with NA values.
• mgidi() to compute the multi-trait genotype-ideotype distance index.
• plot_bars() to create bar plots quickly. Thanks to @MariaDiel for her suggestion.

Minor changes

• Deprecated functions hm_mean() and gm_mean() removed in favour of hmean() and gmean(), respectively.
• Deprecated argument rep retired in Fox(), ge_effects(), Huehn(), resp_surf(), superiority(), and Thennarasu()
• Deprecated argument verbose retired in anova_ind()
• Deprecated argument region retired in resp_surf()
• Remove dependency on dendextend by using ggplot2-based graphics in plot.wsmp().
• Update package site with pkgdown v1.5.0.
• Update documentation in ge_plot()
• Allow using fai_blup() with gamem()
• Improve checking process with inspect()
• Improve feedback for results, indicating random and fixed effects. Thanks to @NelsonJunior for his suggestion.
• plot() call on objects of class gamem, waasb and waas now returns the variable names automatically. Thanks to @MdFarhad for suggesting me this change.
• plot.gamem() and plot.waasb() have a new argument (type = "vcomp") to produce a plot showing the contribution of the variance components to the phenotypic variance
• cv_ammi(), cv_ammif(), and cv_blup() now check for missing values and unbalanced data before computing the cross-validation. (#3)

Bug fixes

• Fix problems from a recent upgrade of package tibble to version 3.0.0.
• get_model_data() now fills rows that don't matches across columns with NA. Thanks to @MdFarhad for his report.
• get_model_data() called now report mean squares, F-calculated and P-values for blocks within replicates in anova_ind().

metan 1.4.0

Bug fixes

• Factor columns can now have custom names rather than ENV, GEN, and REP only (#2).

New functions

• gmd() a shortcut to get_model_data()
• gtb() to generate a genotype-by-trait biplot.
• gamem_met() to analyze genotypes in multi-environment trials using mixed- or random-effect models allowing unbalanced data. Thanks to @EderOliveira for his e-mail.
• has_class() to check if a class exists.
• impute_missing_val() to impute missing values in a two-way table based on Expectation-Maximization algoritms.
• non_collinear_vars() to select a set of predictors with minimal multicollinearity.
• replace_na() to replace NA values quicly.
• random_na() to generate random NA values based on a desired proportion.

Minor changes

• gge(), performs_ammi(), waas(), and waasb() now handle with unbalanced data by implementing a low-rank matrix approximation using singular value decomposition to impute missing entires. Imputation generates a warning message.
• NA values are checked and removed with a warning when computing stability indexes. Thanks to @MdFarhad for alerting me.
• New argument plot_res in path_coeff() to create a residual plot of the multiple regression model.
• Update the citation file to include the published official reference.
• Argument verbose deprecated in functions anova_ind() and split_factors()
• Argument rep deprecated in functions Fox(), Huehn(), superiority(), and Thennarasu().
• Deprecated argument means_by removed in functions can_corr() and clustering().
• Deprecated argument verbose removed in functions colindiag() and split_factors().
• Deprecated argument values removed in functions desc_stat() and find_outliers().
• Deprecated argument var removed in function desc_wider().
• Remove dependency on lattice by using ggplot2 in plot.resp_surf().
• An up-to-date cheat sheet was included.

metan 1.3.0

New functions

• alpha_color() To get a semi-transparent color
• gafem() To analyze genotypes using fixed-effect models.
• residual_plots() A helper function to create residuals plots.
• stars_pval() To generate significance stars from p-values
• doo() An alternative to dplyr::do for doing anything

utils_stats

• cv_by() For computing coefficient of variation by levels of a factor.
• max_by() For computing maximum values by levels of a factor.
• means_by() For computing arithmetic means by levels of a factor.
• min_by() For computing minimum values by levels of a factor.
• n_by() For getting the length.
• sd_by() For computing sample standard deviation.
• sem_by() For computing standard error of the mean by levels of a factor.
• av_dev() computes the average absolute deviation.
• ci_mean() computes the confidence interval for the mean.
• cv() computes the coefficient of variation.
• hm_mean(), gm_mean() computes the harmonic and geometric means, respectively. The harmonic mean is the reciprocal of the arithmetic mean of the reciprocals. The geometric mean is the nth root of n products.
• kurt() computes the kurtosis like used in SAS and SPSS.
• range_data() Computes the range of the values.
• sd_amo(), sd_pop() Computes sample and populational standard deviation, respectively.
• sem() computes the standard error of the mean.
• skew() computes the skewness like used in SAS and SPSS.
• sum_dev() computes the sum of the absolute deviations.
• sum_sq_dev() computes the sum of the squared deviations.
• var_amo(), var_pop() computes sample and populational variance.
• valid_n() Return the valid (not NA) length of a data.

utils_rows_cols

• colnames_to_lower(), colnames_to_upper(), and colnames_to_title() to translate column names to lower, upper and title cases quickly.

utils_num_str

• all_lower_case(), all_upper_case(), and all_title_case() to translate strings vectors or character columns of a data frame to lower, upper and title cases, respectively.
• tidy_strings() Tidy up characters strings, non-numeric columns, or any selected columns in a data frame by putting all word in upper case, replacing any space, tabulation, punctuation characters by '_', and putting '_' between lower and upper cases.
• find_text_in_num() Find text fragments in columns assumed to be numeric. This is especially useful when everything() is used in argument resp to select the response variables.

New arguments

• anova_ind(), anova_joint(), performs_ammi(), waas() and waasb(), now have the argument block to analyze data from trials conducted in an alpha-lattice design. Thanks to @myaseen208 for his suggestion regarding multi-environment trial analysis with alpha-lattice designs.
• argument repel included in plot_scores() to control wheater the labels are repelled or not to avoid overlapping.

Deprecated arguments

Argument means_by was deprecated in functions can_corr() and clustering(). Use means_by() to pass data based on means of factor to these functions.

Minor changes

• Change "#000000FF" with "#FFFFFF00" in transparent_color()
• desc_stat() now handles grouped data passed from dplyr::group_by()
• plot_scores() now support objects of class waas_mean.
• Include inst/CITATION to return a reference paper with citation("metan").
• Change 'PC2' with 'PC1' in y-axis of plot_scores(type = 2) (#1)
• get_model_data() now support models of class anova_joint and gafem and extract random effects of models fitted with waasb() and gamem().
• Update plot.waasb() and plot.gamem() to show distribution of random effects.
• inspect(), cv_blup(), cv_ammif(), and cv_ammi() now generate a warning message saying that is not possible to compute cross-validation procedures in experiments with two replicates only. Thanks to @Vlatko for his email.
• plot.wsmp() now returns heatmaps created with ggplot2. Thus, we removed dependency on gplots.
• Vignettes updated

metan 1.2.1

• References describing the methods implemented in the package were included in description field of DESCRIPTION file as suggested by the CRAN team.

metan 1.2.0

• Minor changes

  • corr_plot() now don't write a warning message to the console by default.
  • select_numeric_cols() now is used as a helper function in metan.
  • metan now reexports mutate() from dplyr package.
  • get_model_data() now set default values for each class of models.
  • Argument by that calls internally split_factors() included to facilitate the application of the functions to each level of one grouping variable.

• New functions

  • add_cols(), and add_rows() for adding columns and rows, respectively.
  • remove_cols(), and remove_rows() for removing columns and rows, respectively.
  • select_cols() and select_rows() for selecting columns and rows, respectively.
  • select_numeric_cols(), and select_non_numeric_cols() for selecting numeric and non-numeric variables quickly.
  • round_cols() for rounding a whole data frame to significant figures.
  • all_lower_case(), and all_upper_case() for handling with cases.
  • extract_number(), extract_string(), remove_strings(), replace_number(), and replace_string(), for handling with numbers and strings.
  • get_level_size(), and get_levels() for getting size of levels and levels of a factor.
  • means_by() for computing means by one or more factors quickly.
  • ge_means() for computing genotype-environment means
  • ge_winners() for getting winner genotypes or ranking genotypes within environments.
  • env_dissimilarity() for computing dissimilarity between test environments.

metan 1.1.2

• Reexport select_helpers starts_with(), ends_with(), contains(), contains(), num_range(), one_of(), everything(), and last_col().
• When possible, argument resp (response variable(s) now support select helpers.
• New helper function sem() for computing standard error of mean.
• New helper functions remove_rows_na() and remove_cols_na() for removing rows or columns with NA values quickly.
• New select helpers difference_var(), intersect_var(), and union_var() for selecting variables that match an expression.
• New function Schmildt() for stability analysis.
• Plot regression slope and mean performance in objects of class ge_reg.
• Update get_model_data() to support objects of class Schmildtand Annicchiarico.

metan 1.1.1

• Now on.exit() is used in S3 generic functions print() to ensure that the settings are reset when a function is excited.
• Computationally intensive parts in vignettes uses pre-computed results.

metan 1.1.0

I'm very pleased to announce the release of metan 1.1.0, This is a minor release with bug fixes and new functions. The most important changes are described below.

• New function corr_stab_ind() for computing Spearman's rank correlation between stability indexes;
• New function corr_coef() for computing correlation coefficients and p-values;
• New S3 method plot.corr_coef() for creating correlation heat maps;
• New S3 method print.corr_coef() for printing correlation and p-values;
• New helper functions make_lower_tri() and make_upper_tri() for creating lower and upper triangular matrices, respectively.
• New helper function reorder_cormat() for reordering a correlation matrix according to the correlation coefficients;
• Improve usability of get_model_data() by supporting new classes of models. Now, get_model_data() can be used to get all statistics or ranks computed with the wrapper function ge_stats().
• arrange_ggplot() now support objects of class ggmatrix.
• Change the default plot theme to theme_metan()
• Update function's documentation;
• Update vignettes.

metan 1.0.2

• New function arrange_ggplot() for arranging ggplot2 graphics;
• New function ge_effects() for computing genotype-environment effects;
• New function gai() for computing the geometric adaptability index;
• New helper function gm_mean() for computing geometric mean;
• New helper function hm_mean() for computing harmonic mean;
• New helper function Huehn() for computing Huehn's stability statistic;
• New helper function Thennasaru() for computing Thennasaru's stability statistic;
• Improve usability of get_model_data() by supporting new classes of models;
• Update function's documentation;
• Update vignettes;

metan 1.0.1

• New function gamem() for analyzing genotypes in one-way trials using mixed-effect models;
• New function desc_wider() to convert an output of the function desc_stat() to a 'wide' format;
• New function Fox() for Stability analysis;
• New function Shukla() for stability analysis;
• New function to_factor() to quickly convert variables to factors;
• Improve usability of get_model_data() function;
• Update function's documentation;
• Update vignettes;

metan 1.0.0

The changes in this version were made based on suggestions received when metan was submitted to CRAN for the first time.

Major changes

The documentation of the following functions was updated by including/updating the \value section of .Rd files.

• AMMI_indexes()
• Annichiarico()
• anova_ind()
• as.lpcor()
• as.split_factors()
• bind_cv()
• can_cor()
• comb_vars()
• corr_ci()
• corr_plot()
• covcor_design()
• cv_ammi()
• cv_ammif()
• cv_blup()
• desc_stat()
• ecovalence()
• fai_blup()
• ge_factanal()
• ge_plot()
• ge_reg()
• ge_stats()
• get_model_data()
• is.lpcorr()
• is.split_factors()
• mahala()
• mahala_design()
• make_mat()
• make_sym()
• mtsi()
• pairs_mantel()
• plot.*() and plot_*() functions
• rbind_fill()
• resca()
• resp_surf()
• waas()
• wsmp()
• waasb()

Minor changes

To allow automatic testing, the examples of the following functions were unwrapped by deleting \dontrun{}.

• bind_cv()
• clustering()
• comb_vars()
• corr_ci()
• corr_plot()
• covcor_design()
• desc_stat()
• ecovalence()
• path_coefff()
• plot.fai_blup()
• plot.mtsi()
• plot.wsmp()
• plot_ci()
• wsmp()

In the examples of the functions for cross-validation \dontrun{} was changed with \donttest{}

• cv_ammi()
• cv_ammif()
• cv_blup()
• plot.cv_ammif()

metan 0.2.0

This is the first version that will be submitted to CRAN. In this version, deprecated functions in the last versions were defunct. Some new features were implemented.

• New functions

  • fai_blup() computes the FAI-BLUP index (https://onlinelibrary.wiley.com/doi/full/10.1111/gcbb.12443)
  • gge() computes the genotype plus genotype-vs-environment model.
  • plot_factbars() and plot_factlines() are used to create bar and line plots, respectively, considering an one- or two-way experimental design.
  • desc_stat() computes several descriptive statistics.
  • can_corr()computes canonical correlation coefficients.
  • resp_surf() computes response surface model using two quantitative factors.
  • make_mat() is used to create a two-way table using two columns (factors) and one response variable.
  • make_sym() is used to create a symmetric matrix using a upper- or lower-diagonal matrix.

• Minor improvements

  • New evaluation for text vectors are now used in the functions AMMI_indexes() and fai_blup() and desc_stat(). For example, to indicate the statistics to be computed in desc_stat() you must use now stats = c("mean, SE.mean, CV, max, min")) instead stats = c("mean", "SE.mean", "CV", "max", "min"))

metan 0.1.5

In the latest development version, the package METAAB was renamed to metan (multi-environment trials analysis). Aiming at a cleaner coding, in this version, some functions were deprecated and will be defunct in the near future. Alternative functions were implemented.

• For WAAS.AMMI(), use waas().
• For WAASBYratio(), use wsmp().
• For WAASratio.AMMI(), use wsmp().
• For autoplot.WAAS.AMMI(), use autoplot.waas().
• For plot.WAASBYratio(), use plot.wsmp().
• For plot.WAASratio.AMMI(), use plot.wsmp().
• For predict.WAAS.AMMI(), use predict.waas().
• For summary.WAAS.AMMI(), use summary.waas()

Widely-known parametric and nonparametric methods were implemented, using the following functions.

• Annicchiarico() to compute the genotypic confidence index.
• ecovalence() to compute the Wricke's ecovalence.
• ge_factanal() to compute to compute the stability and environmental.
• ge_reg() to compute the joint-regression analysis. stratification using factor analysis.
• superiority() to compute the nonparametric superiority index.

METAAB 0.1.4

In the latest development version, some useful functions were included. One of the most interesting features included in this version was allowing the functions to receive data from the forward-pipe operator %>%. Bellow are the functions included in this version.

• anova_ind() to perform a within-environment analysis of variance easily;
• colindiag() to perform a collinearity diagnostic of a set of predictors;a
• find_outliers() to easily find possible outliers in the dataset;
• group_factors() to split a dataset into a list of subsets using one or more grouping factors. This function may be used befor some functions, e.g., find_outliers(), colindiag(), path_coeff() to compute the statistics for each level of the factor (or combination of levels of factors).
• lpcor() to compute linear and partial correlation coefficients.
• pairs_mantel() to compute a graphic pairwise Mantel's test using a set of correlation matrices;
• path_coeff() to compute path coefficients with minimal multicollinearity;

The following S3 Methods were also implemented:

• is.group_factors() and as.group_factors() to check or easily coerce a dataframe that has one or more factor columns to an object of group_factors;
• is.lpcorr() and as.lpcorr() to check or easily coerce a list of correlation matrices to an object of lpcorr;

METAAB 0.1.3

• AMMI-based stability indexes;
• Allow analyzing multiple variables at the same time;
• S3 methods such as plot(), predict(), summary() implemented.

METAAB 0.1.2

• Mixed-effect model with environment random effect;
• Random-effect model.

METAAB 0.1.1

• The first version of the package