Release 12: new features

1. Highlights

●  produced in 2009

●  2 new directives, 42 new procedures and 41 new functions

●  regression – quantile regression(RQLINEAR and RQSMOOTH), zero inflated Poisson models (R0INFLATED), parameter confidence limits (FIT etc)

●  ANOVA – model terms as covariates (AFCOVARIATES), trellis plots of means (AGRAPH)

●  REML – Akaike and Schwartz information coefficients (VAIC), contrasts (FCONTRASTS) and diallel models (FDIALLEL)

●  design – concurrence matrices (FOCCURRENCES), complements of incomplete block designs (FCOMPLEMENT), link to CycDesigN

●  MVA – permutation tests in MANOVA, biplot enhancements (DBIPLOT)

●  variety trials – stability analysis (GESTABILITY), G + G.E biplots (GGEBIPLOT)

●  QTL estimation by REML – DQMAP, DQMKSCORES, DQMQTLSCAN, DQSQTLSCAN, QCANDIDATES, QDESCRIBE, QEXPORT, QIBDPROBABILITIES, QIMPORT, QMBACKSELECT, QMESTIMATE, QMQTLSCAN, QSBACKSELECT, QSESTIMATE, QSQTLSCAN, QTHRESHOLD and VGESELECT)

●  zero-inflated regression models for count data with excess zeros (R0INFLATED and R0KEEP)

●  hot-deck and model-based imputation for survey data (SVHOTDECK)

●  ability to fix or initialize the dispersion factors of the additional random terms in HGLMs and DGHLMs (HGFIXEDMODEL, HGRANDOMMODEL, HGDRANDOMMODEL)

●  additional links to the NAG Library (NAG)

●  graphics – oblique axes (AXIS), interactive tools (hotpoints to show/hide items; DGRAPH and D3GRAPH)

●  image functions – 33 new functions for image manipulation

●  64-bit implementation(s)

●  additional documentation

2. What’s new

2.1 Directives

AXIS defines an oblique axis for high-resolution graphics.

FRQUANTILES forms regression quantiles.

2.2 Procedures

AFCOVARIATES defines covariates from a model formula for ANOVA.

ALLDIFFERENCES shows all pairwise differences of values in a variate or table.

DBIPLOT plots a biplot from an analysis by PCP, CVA or PCO.

DCORRELATION plots a correlation matrix.

DFRTEXT adds text to a graphics frame.

DFUNCTION plots a function.

DHSCATTERGRAM plots an h-scattergram.

DQMAP displays a genetic map.

DQMKSCORES plots a grid of marker scores for genotypes and indicates missing data.

DQMQTLSCAN plots the results of a genome-wide scan for QTL effects in multi-environment trials.

DQSQTLSCAN plots the results of a genome-wide scan for QTL effects in single-environment trials.

FCOMPLEMENT forms the complement of an incomplete block design.

FCONTRASTS modifies a model formula to contain contrasts of factors.

FDIALLEL forms the components of a diallel model for REML or regression.

FOCCURRENCES counts how often each pair of treatments occurs in the same block.

GESTABILITY calculates stability coefficients for genotype-by-environment data.

GGEBIPLOT plots displays to assess genotype+genotype-by-environment variation.

HGSTATUS displays the current HGLM model definitions.

MAREGRESSION does regressions for single-channel microarray data.

MC1PSTATIONARY gives the stationary probabilities for a 1st-order Markov chain.

MINIMIZE finds the minimum of a function calculated by a procedure.

MOVINGAVERAGE calculates and plots the moving average of a time series.

QCANDIDATES selects QTLs on the basis of a test statistic profile along the genome.

QDESCRIBE calculates descriptive statistics of molecular markers.

QEXPORT exports genotypic and phenotypic data for QTL analysis.

QIBDPROBABILITIES reads molecular marker data and calculates IBD probabilities.

QIMPORT imports genotypic and phenotypic data for QTL analysis.

QMBACKSELECT performs a QTL backward selection for loci in multi-environment trials.

QMESTIMATE calculates QTL effects in multi-environment trials.

QMQTLSCAN performs a genome-wide scan for QTL effects (Simple and Composite Interval Mapping) in multi-environment trials.

QSBACKSELECT performs a QTL backward selection for loci in single-environment trials.

QSESTIMATE calculates QTL effects in single-environment trials.

QSQTLSCAN performs a genome-wide scan for QTL effects (Simple and Composite Interval Mapping) in single-environment trials.

QTHRESHOLD calculates a threshold to identify a significant QTL.

RQLINEAR fits and plots quantile regressions for linear models.

RQSMOOTH fits and plots quantile regressions for loess or spline models.

R0INFLATED fits zero-inflated regression models to count data with excess zeros.

R0KEEP saves information from a zero-inflated regression model for count data with excess zeros fitted by R0INFLATED.

SETDEVICE opens a graphical file and specifies the device number on basis of its extension.

SVHOTDECK performs hot-deck and model-based imputation for survey data.

VAIC calculates the Akaike and Schwarz information coefficients for REML.

VGESELECT selects the best variance-covariance model for a set of environments.

2.2 Functions

BIN	Bessel function In
BJN	Bessel function Jn
BK0	Bessel function K0
BK1	Bessel function K1
BKN	Bessel function Kn
BY0	Bessel function Y0
BY1	Bessel function Y1
BYN	Bessel function Yn
IMBEQUALIZE	Histogram equalization of brightness
IMBLUR	Blur
IMBRIGHTNESS	Modify brightness
IMBSTRETCH	Histogram stretch of brightness
IMCEQUALIZE	Histogram equalization of individual colours
IMCONTRAST	Modify contrast and brightness
IMCREPLACE	Replace one colour with another
IMCSTRETCH	Histogram stretch of individual colours
IMDESPECKLE	Despeckle
IMELLIPSE	Draw ellipse
IMEMBOSS	Emboss
IMGAMMA	Apply gamma correction to brightness
IMGBLUR	Gaussian blur
IMGRAYSCALE or IMGREYSCALE	Convert to grey scale
IMHFLIP	Horizontal flip
IMLINE	Draw line
IMMCONVOLUTION	Apply convolution filter
IMMEDIANFILTER	Median filter
IMOVERLAY	Overlay images
IMPUSH	Point-to-point warp
IMRECTANGLE	Draw coloured rectangle
IMROTATE	Rotate
IMSATURATE	Adjust saturation
IMSHARPEN	Sharpen
IMSIZE	Change size
IMSSTRETCH	Histogram stretch of saturation
IMSTEXT	Draw smooth text
IMTEXT	Draw text
IMUNSHARPEN	Apply unsharp mask
IMVFLIP	Vertical flip
IMXSHEAR	Shear in x direction
IMYSHEAR	Shear in y direction
IM3CONVOLUTION	Apply 3×3 convolution filter

3. What’s changed

You can now use negative numbers in a qualified identifier to exclude units. Hexadecimal colours are now allowed in every option or parameter that allows strings to be used to define colour names (see e.g. PEN). These will be defined in a string of the form '#abc', '0xabc' or '0Xabc', where a b and c are hexadecimal numbers (i.e. numbers 0-9 or letters a-f).

Most of the changes are compatible with Release 11, the previous release. There are a few commands, however, where new options or parameters have been inserted into the existing lists. These may cause problems in statements where option or parameter names have been omitted or abbreviated (see Section 1.7.1 of Part 1 of the Guide to the GenStat Command Language for details). To avoid any difficulty, the name of the option/parameter after the new option/parameter should be given explicitly, and not abbreviated to fewer than four characters.

Any command, where changes in Release 12 may cause incompatibilities in existing programs, is marked in Sections 3.1 and 3.2 by the symbol ^†. The full details are given in Section 3.3.

3.1 Directives

^†AKEEP can save the setting of the FACTORIAL option used in the ANOVA command that performed the analysis and the identifier of the y-variate.

CALCULATE allows a temporary restruction to be defined on vectors, overiding any existing restrictions.

CLUSTER allows the best clustering to be selected using several random starts.

COVARIATE allows you to group covariates together in a pointer if you want their sums of squares to be combined into a single line in the analysis-of-variance table.

CVA, PCO and PCP now have a SAVE option that can save a multivariate save structure containing all the details of the analysis. If SAVE is unset, an unnamed structure is formed and saved automatically (this can then be accessed using the GET directive). Alternatively, you can set SAVE=* if you do not want to form a save structure. This might be useful, for example, if you have a very large data set and do not want to commit the storage space.

DGRAPH and D3GRAPH have a new setting SCREEN=resize to keep the screen but resize the axes to take account of the information that is being added to the plot.

DGRAPH allows you to control whether or not each component of the graph is displayed initially when the graph is plotted. Components can be associated with “hot” points on the graph. Their appearance can then be controlled by the user making right-mouse clicks at the relevant “hot” points.

FACROTATE allows you to control how many dimensions to rotate, and to save the rotation matrix.

^†FIT, ADD, DROP, SWITCH, TRY, STEP and ^†RDISPLAY can print confidence limits for the parameter estimates.

^†FRAME allows you to include oblique axes (defined by AXIS) in a window, and to set the background colour of the entire frame.

FSIMILARITY can calculate similarities for attributes defined using factors.

^†MDS allows the best solution to be selected using several random starts.

^†PEN provides better control for the plotting of textual labels (size, location etc.) and for the size of symbols, and has a new parameter CAREA to specify colours of areas inside polygons and bars of histograms.

^†PRINT can control whether the identifiers are printed for row and column labels.

^†REML can save an exit code to indicate whether the analysis was successful. Another change is that REML now uses a special vsave structure as its save structure, rather than an ordinary pointer.

RESTRICT has a new option NULL to indicate whether a null subset has been defined; when NULL is set, a null subset will not generate a fault or warning (it is assumed that users will then guard against the effects of the null set for themselves).

^†VKEEP can now save the numbers of degrees of freedom fitted in the fixed and random model, the unit × unit variance-covariance matrix, Wald statistics with numerator and denominator numbers of degrees of freedom, the WORKSPACE setting from the REML command, and an exit code to indicate whether the analysis was successful.

^†XAXIS, ^†YAXIS and ^†ZAXIS allow date formats to be defined for axis labels; there is also a new option REVERSE to reverse the direction of the axis.

3.2 Procedures

AFIELDRESIDUALS procedure now allows titles and control of pens for the contour and shade plots.

^†AGRAPH allows tables of means with more than two classifying factors to be plotted e.g. in a trellis plot, and allows more control of the way in which predicted means are formed from unbalanced analyses. You can also specify your own pens to control the symbols, colours and line styles used in a high-resolution graph.

CONVEXHULL allows a scaling factor for the hulls.

^†DDENDROGRAM can now split dendrograms over several pages in a high-resolution plot.

DISCRIMINATE can now save canonical correlation coefficients, within-group correlation matrices of the input variates and within-group correlations between the input and canonical variates.

FACPRODUCT allows you to control whether or not the identifiers of the factors are included in the labels of the new (product) factor.

^†FFRAME has new options to improve the specification of windows for trellis plots.

FITINDIVIDUALLY, FITMULTINOMIAL and ^†RQUADRATIC can print confidence limits for the parameter estimates.

GLMM allows you to set the amount of workspace available for the REML algorithm to use when fitting the model.

HGFIXEDMODEL, HGRANDOMMODEL, ^†HGDRANDOMMODEL now allow you to fix or initialize the dispersion factors of the additional random terms in HGLMs and DGHLMs, and the facilities for dispersion modelling in DHGLMs have been updated to match those in ordinary HGLMs.

MANOVA can perform permutation tests, and allows you to select which test statistics to print.

PROBITANALYSIS parameter NBINOMIAL can now be set to a scalar, and option SEPARATE has a setting notintercept that you can use to fit a common intercept over groups.

^†RJOINT can now plot the model; you can also save the residual deviance and degrees of freedom, and an exit code to indicate whether the model was fitted successfully.

RSCREEN allows you to decide to use all the units available when constructing each marginal or conditional test i.e. to ignore missing values in any explanatory variable that is not involved in the test. (This provides more information for each test, but the tables of tests should be interpreted with care as different tests may be based on different sets of units.)

^†RSEARCH can save the estimates and standard errors for all possible regression models.

R2LINES can include additional x-variates in the model.

Finally, procedure LIBHELP now connects to the on-line help system, instead of the earlier help system that accessed the information in text structures from a backing-store file. The earlier system has been withdrawn, together with procedures FLIBHELP, LIBINFORM and LIBMANUAL. Help information for user procedure libraries can be supplied in GenStat for Windows by putting a Windows (.chm) help file, together with the procedure library (.glb) file, in the add-in folder. The Windows Help menu is then extended to contain a sub-option with the name of the procedure library, in its User Libraries option. Clicking on that sub-option opens the file at its contents page. The names of the procedures can be added to the context-sensitive help by including a topic in the help file for each procedure, and naming these by the procedure names (in full). Users can then access the help for a procedure by placing the curser in the name of the procedure in GenStat for Windows‘ Output window, and pressing the F1 key (in the same way as for the standard GenStat commands, functions and terminology).

3.3 Functions

The existing functions are all unchanged.

3.4 Incompatibilities

AKEEP directive	option AFACTORIAL inserted before WEIGHTS, and YVARIATE inserted before AOVTABLE.
AGRAPH procedure	options COMBINATIONS and ADJUSTMENT inserted before LSDLEVEL, parameter TRELLISGROUPS inserted before MEANS.
CRBIPLOT	option KEYWINDOW inserted before SAVE.
CRTRIPLOT	option KEYWINDOW inserted before SAVE.
DDENDROGRAM	options NPAGES and PAGEINFORMATION inserted before ENDACTION.
HGDRANDOMMODEL procedure	option DLINK inserted before DFORMULA, and DOFFSET inserted before LMATRIX.
FIT directive	option PROBABILITY inserted before NGRIDLINES.
DAPLOT procedure	was replaced by extensions to APLOT in Release 6, but it was retained for compatibility with previous programs. It has now been removed from the documentation, and the procedure itself may be removed in a future release.
FFRAME procedure	options RMLOWER, RMUPPER, CMLOWER and CMUPPER inserted before BACKGROUND.
FRAME directive	option BACKGROUND inserted before RESET, and parameter AXES inserted before SAVE.
FSPREADSHEET procedure	option DUMMY renamed ADUMMY; option READONLY now controls the read-only status only of the whole spreadsheet, with settings no and yes; new parameter PROTECT to control the read-only status of individual columns
MDS directive	NSTARTS and SEED options inserted before MAXCYCLE.
LIBINFORM procedure	has been withdrawn (the information is all easily available in the on-line help)
LIBMANUAL procedure	has also been withdrawn (the procedure-library manual is available in a pdf file accessible from the Help menu)
NOTICE procedure	PRINT setting library has been renamed release, and gives information about the new release as a whole, not just the procedure library.
PEN directive	parameters CAREA, SMSYMBOL and SMLABEL inserted before DFSPLINE; YLSIZE, XLSIZE, YLOFFSET and XLOFFSET inserted before BARTHICKNESS.
PRINT directive	options RLPRINT and CLPRINT have a new setting identifiers to control whether the identifiers are printed for row and column labels. The labels setting now prints only the labels, but the defaults are changed to labels, identifiers to provide the same default output as in previous releases.
RDISPLAY directive	option PROBABILITY inserted before SAVE.
REML directive	parameter EXIT inserted before SAVE. REML now uses a special vsave structure as its save structure, rather than an ordinary pointer. This will require changes if you have a procedure that takes a REML save structure as a setting of one of its options or parameters and you have previously been defining the type as pointer (using the TYPE parameter of OPTION or PARAMETER).
REPLICATION procedure	was replaced by ASAMPLESIZE at Release 8, but has been retained for compatibility with previous programs. It has now been removed from the documentation, and the procedure itself may be removed in a future release.
RJOINT procedure	options TITLE, YTITLE and XTITLE inserted before TOLERANCE.
RQUADRATIC directive	option PROBABILITY inserted before STATIONARY.
RSEARCH directive	options ESTIMATES and SE inserted before RESULTS.
SET directive	option MSAVE inserted before RSAVE.
VKEEP directive	options UVCOVARIANCE, DFFIXED, DFRANDOM, FMETHOD, WORKSPACE and EXIT inserted before SAVE.
XAXIS directive	option DREPRESENTATION inserted before YOMETHOD; option REVERSE inserted before SAVE; LOWER values greater than UPPER values are now faulted (option REVERSE should be used to reverse the direction of the axis instead).
YAXIS and ZAXIS directives	option DREPRESENTATION inserted before XOMETHOD; option REVERSE inserted before SAVE; LOWER values greater than UPPER values are now faulted (option REVERSE should be used to reverse the direction of the axis instead).