Skip to contents

Changelog

Source: NEWS.md

DiallelAnalysisR 1.0.0

Changes visible to users

  • Every function keeps its name, its arguments and the names of its results, so existing scripts run unchanged.
  • Results print, summarise and plot: plot() draws the gca effects with standard-error bars and the sca effects as a heat map (Griffing(), PartialDiallel()), and the Wr-Vr graph (Hayman()); any component converts to a table with data.table::as.data.table(x, component = "ANOVA").
  • The usual model functions work on every result: summary() (t tests of every effect, the crosses with the largest sca effects), anova(), coef(), vcov(), confint(), fitted() and residuals() (plot level), nobs(), df.residual(), predict() (crosses from gca alone or from the full model) and ggplot2::autoplot(); plot(which = "residuals") and plot(which = "qq") check the block analysis.
  • Column arguments accept a bare name (y = Yield), a string (y = "Yield") or a vector.
  • In Methods 2 and 4 a cross may be recorded either way round; missing crosses are reported by name.

Corrections

  • Griffing(), Model 1: the standard errors now use the error mean square; version 0.6.0 used the replicate mean square (with GriffingData1, Method 1: standard error of a gca effect 1.25 instead of 2.17).
  • Griffing(), Method 2: the standard errors of the sca effects of selfs and of crosses were swapped.
  • Griffing(), Method 4: the third standard error (a Method 1 formula that has no meaning in Method 4) is NA; the table of means has every parent in its rows and columns (was 7 x 7 with misaligned labels).
  • Griffing(), Model 2, Methods 1 and 3: the p-value of the reciprocal line uses the error degrees of freedom (was the sca degrees of freedom).
  • Hayman(): the column arguments are used (version 0.6.0 read the columns Cross1 and Cross2 by name).
  • Hayman(): the lines b, b3 and d of the analysis of variance divided by the number of replicates where the formula has 4; the two agree only with 4 replicates (as in HaymanData).
  • Hayman(): H2 = 4 V1L1 - 4 V0L1 - 2E (version 0.6.0 added 4 V0L1); Wr and Vr come from the reciprocal-averaged table (version 0.6.0 took Vr from the upper triangle and Wr from the unaveraged rows). F, H1 and the derived ratios change accordingly.
  • PartialDiallel(): the column arguments are used (version 0.6.0 read y, Rep, Cross1, Cross2 by name); the error line is on the plot scale (version 0.6.0 divided it by the number of replicates, inflating every F); the components are variance components (version 0.6.0 gave differences of mean squares).

New

  • Griffing(), Model 1: new components GCA, SCA (selfs included), Reciprocal; all results: RCBD (the block analysis of the plot data). Hayman(): new component Tests (uniformity t2, b = 0, b = 1); named derived ratios.
  • Griffing() handles missing plots (not missing crosses): least-squares cross means adjusted for blocks, with the harmonic mean of the plot numbers as the number of replicates.
  • GriffingMeans(): Griffing’s analysis of a table of cross means with the error mean square and degrees of freedom of the experiment; data set Griffing56 (Griffing 1956, Table 6) and 77 published values of Griffing’s Tables 8, 9, 10 and 13 in reference.csv, recomputed by da_reference().
  • Heterosis(): mid-parent, better-parent and standard heterosis with t tests.
  • Griffing(..., Reciprocal = "split"): maternal and non-maternal reciprocal effects (Cockerham and Weir 1977) in Methods 1 and 3, with their tests, components and standard errors.
  • GardnerEberhart(): Gardner and Eberhart’s Analyses II and III.
  • LineTester(): line by tester analysis with proportional contributions, effects, standard errors and genetic variances; NCDesign(): North Carolina designs I, II and III; data sets LineTesterData and NCData.
  • GriffingEnv(): Griffing’s analysis over several environments with gca, sca and reciprocal by environment interactions and, in Model 2, their variance components.
  • plot(x, which = "biplot") for Griffing() and LineTester() results: a biplot of the cross means.
  • DiallelSim() simulates diallel data with known effects; DiallelPower() estimates the power of the combining-ability tests for numbers of replicates.
  • da_tables() and da_export(): the main tables of any result in Word, LaTeX, HTML, Markdown or CSV.
  • Title and description of the package cover the new analyses; citation("DiallelAnalysisR") gives the reference of the package (also in the README and on the site).
  • DiallelApp(): a Shiny application for every analysis, with plots, exports and the R code of the analysis.
  • Complete help pages with the models and formulas; fifteen vignettes (HTML and PDF); notation table da_notation(); plot theme da_theme() and da_save().
  • Tests of every analysis against independent calculations and against version 0.6.0.

Internal

  • One computational core per analysis instead of eight copied branches; closed forms throughout; up to 60 times faster for 20 parents.

DiallelAnalysisR 0.6.0

CRAN release: 2024-09-13

  • Updated the DESCRIPTION file to fulfil the requirements of CRAN.

DiallelAnalysisR 0.5.0

CRAN release: 2023-03-31

  • Refined code.

DiallelAnalysisR 0.4.0

CRAN release: 2021-02-25

  • Removed vignettes.

DiallelAnalysisR 0.3.0

CRAN release: 2020-08-01

  • Added PartialDiallel() for the analysis of partial diallels.
  • Added vignettes; refined code and documentation.

DiallelAnalysisR 0.1.1

CRAN release: 2016-04-09

  • First version: Griffing’s and Hayman’s diallel analyses.