4.2 General rules

The parameter file is read line by line. Each line is assumed to be 88 characters long, i.e. anything in columns 89 onwards is ignored.

Any line with a # in column 1 is considered to be a comment line and is skipped.

WOMBAT relies on specific codes at the beginning of each line (leading blank spaces are ignored) to distinguish between various types of information given. Valid codes are summarised in 4.1. Most codes can be abbreviated to 3 letters. The code and the information following it should be separated by space(s). Codes are not case sensitive, i.e. can be given in either upper or lower case letters.

Depending on the initial code, WOMBAT expects further information on the same or following lines. All variable and file names specified are treated as case sensitive. File names can be up to 30 characters long, and variable names can comprise up to 20 characters. All codes, names and variables must be separated by spaces.

The parameter file can have two different types of entries :

1. ‘Short’ entries (e.g. file names, analysis type, comment) which consist of a single line.
Each of these lines must start with a code for the type of information given.

EXAMPLE:

PEDS pedigrees.dat

Here PEDS is the code for pedigree information, “pedigrees.dat” is the name of the file from which pedigree information is to be read.


2. ‘Long’ or block entries, spanning several lines (e.g. for the layout of the data file, or the model of analysis).
Each of these entries starts with a line which gives the code for the entry (see 4.1), and possibly some additional information. This is followed by lines with specific information, where of each of these lines again starts with a specific code. Except for blocks of starting values of covariance components, the block is terminated by a line beginning with END.

EXAMPLE:

MODEL
  FIX  CGroup
  RAN  Animal  NRM
  TRA  Weight
END MODEL

This shows a block entry for the model of analysis, where the model fits CGroup as a crossclassified, fixed effect and Animal as random effect with covariance matrix proportional to the numerator relationship matrix, and Weight is the trait to be analysed.


Different entries should be specified in the order in which they are listed in the following sections.


Table 4.1: Valid codes for entries in the parameter file
Code Within Indicator for
RUNOP Line with run options
COMMENT  Comment line
PED S Name of pedigree file
DATA Name of data file
TRNOS DAT Trait numbers (grouped input)
NAMES DAT Multiple column names (grouped input)
ANALYSIS Code for analysis type
MODEL Model of analysis
FIX MOD Name of fixed effect
COV MOD Name of fixed covariable
RAN MOD Name of random effect
RRC MOD Name of control variable
SUBJ MOD Name of variable identifying “subject”
EXT MOD Name extra variable
TRAIT MOD Trait name
ZEROUT Fixed effects levels to be set to zero
PSEUDOFX Random effects levels to be treated as random
SE+USR Additional, user defined functions of covariances
SUM SE+ Weighted sum
VRA SE+ Variance ratio
COR SE+ Correlation
SPECIAL Miscellaneous options
COVZER SPECIAL How to deal with ‘zero’ covariables
WEIGHT SPECIAL Weighted analysis
REPEAT SPECIAL Run with low proportion of repeated records
CLONES SPECIAL Ignore sorting of data file
VARIANCE Name of random effect
RESIDUAL
ERROR
END