Echidna.options: Model Qualifiers for Echidna
In yzhlinscau/AAfun0s: Added Functions for Echidna

Echidna.options

R Documentation

Model Qualifiers for Echidna

Model Qualifiers for Echidna

The Qualifiers control various aspects of the model fitting process and reporting of results:

jobqualf	Description
`!WORKSPACE n`	where n is an integer between 1 and 32 gigabytes.
`!CONTINUE [f] or !FINAL [f]`	instructs Echidna to retrieve variance parameters from an .esv file.
`!VIEW`	displays any Winteracter graphics directly to the screen, such as '!view !PNG'
`!DEBUG`	requests additional debugging output to be written.
`!LOGFILE`	receive the DEBUG information.
`!EQN [q]`	sets the !EQN qualifier which selects the equation ordering option, q could be 1~7 or 77.
`others`	!RENAME,!ARGS,!OUTFOLDER, no needs in R.

qualifier	Description
`!EXTRA n`	forces n more iterations after convergence .
`!SINGLE`	forces Echidna not to use Parallel Processing.
`!SLOW`	reduces stepsize when updating variance parameters so convergence is slower but possible more reliable.
`!SKIP i`	indicating the file contains i heading lines to be ignored.
`!READ f`	specifying that f data fields are to be read.
`!FILTER Variable !SELECT value !EXCLUDE value`	subset the data sets.
`!MVINCLUDE`	design variables which are missing are assumed to be zero.
`!MVREMOVE`	data records are ignored if any design variables have missing values.

GRM qualifiers	Description
`!SKIP i`	to skip the first i lines of the grm/giv file.
`!LDET d`	lets you set the logDet value for the supplied giv matrix.
`!ADD value`	adds value to the diagonal of a GRM matrix before inversion.value can only be 1, 2 or 3.
`!PSD`	allows a GRM matrix to be positive semidefinite.
`!NSD`	allows a GRM matrix to be negative semidefinite.
`!ND`	allows a GRM matrix to be negative definite.

Data fields: if VARIABLE is a CLASS variable, the NAME must be followed by a qualifier indicating it is a class variable.

class qualifier	Description
`*`	variable is coded 1:n.
`!I n`	variable is coded with INTEGER labels, not 1:n.
`!A n`	variable is coded with ALPHANUMERIC labels.
`!L <labels>`	data is coded 1:n and <labels> is the list of n class names.
`!AS <factor>`	when two or more alphanumeric variables have a common list of class names.
`!LL c`	reset the maximum length to c characters.
`!PRUNE`	reset the number of levels (n) in a factor to the real maximum levels.

PEDIGREE FILE: ifile contains 3 or 4 fields being the identity (tag, id, name) of an individual, its Sire and its Dam..

qualifier	Description
`!SKIP i`	the file contains i heading lines to be ignored.
`!CSV`	file is strictly COMMA delimited, without .csv suffix.
`!GROUPS g`	first g lines are genetic groups. .
`!MGS`	the third field is a maternal grandsire (not DAM).
`!SAVE f`	requests the inverse be written as a .giv (f=1, ascii) or .bgiv (f=3, real binary) file.

VPREDICT statements have the syntax as: <Key Letter> <Label> <arguments>. <Key Letter> is one of the letters F, H, R, V, etc.

Key Letter	Description
`F`	specifies a linear function like animal + units.
`H`	specifies a ratio (heritability) like animal / Total.
`R`	specifies correlations from a matrix or formula.
`V`	converts a FA structure to a US structure.
`X`	is a multiply function.
`S`	is a square root function.
`K`	sets short vectors for constants (e.g. Legendre coefficients) to be used for M.
`M`	uses the K coefficients to convert US (us(leg(dim,3))) matrix for a specific dim.
`C`	moves components back e.g. C label I[:II]=J[:JJ] where I < J.
`D`	discards components eg D from 393.
`W`	writes components to f.vpc and their variance to f.vpv.

Variance structures

function	Description
`id(),idv()`	identity or scaled identity.
`ar1(), ar1v()`	autoregressive correlation or covariance matrix.
`coru(), coruv(), coruh()`	uniform correlation matrix, v for common variance,h for heterogeneous variance.
`diag()`	diagonal variance matrix.
`grmk()`	the kth GRM matrix.
`mrmk()`	multiple relationship matrices.
`us()`	unstructured variance matrix.
`facvk()`	factor analytic (basic form).
`xfak()`	factor analytic (eXtended form).
`rrk()`	factor analytic: No specific variance.