11.2 Sample parameter files for lm_lods, lm_markers, lm_bayes and lm_schnell

Since lm_markers is our preferred program and it can analyze genotypic, discrete and quantitative traits, we will focus on the parameters files for this program. In the example package, under the `Lodscores' subdirectory, these three files run with lm_markers: `ped73_ge.par', `ped73_ph.par' and `ped73_qu.par'.

Except for the statements specifying the trait, the three parameter files share the following statements:

input pedigree file '../ped73.ped' input seed file '../sampler.seed' output overwrite seed file '../sampler.seed' input marker data file '../ped73.marker.missing' select all markers traits 1 map trait 1 all interval proportions 0.3 0.7 map trait 1 external recomb fracts 0.05 0.15 0.3 0.4 0.45 set trait 1 freqs 0.5 0.5 sample by scan set burn-in iterations 150 set L-sampler probability 0.2 set MC iterations 3000 check progress MC iterations 1000

The pedigree file specified by the `input pedigree file' statement can contain multiple traits. Additional statements (see below) are needed to specify the type of trait and the column number in the pedigree file. The marker map, allele frequencies and genotypes are given in a separate file `ped73.marker.missing', included using the `input marker data file' statement. Genotypes, missing at random at some markers for some individuals, are for only observed individuals in `ped73'. Alternatively, as demonstrated in the parameter file `ped73_qu.par', you may include marker information using statements `map marker dist' and `set markers ... freqs', leaving only marker genotypes in a separate input file.

The two `map trait' statements give trait locus positions at which the LOD scores should be calculated, for intervals between marker loci and for locations outside the range of marker loci. Between two marker loci, the positions are specified by proportions, at 30 and 70 percent of the interval (it can deal with gender--specific maps easily this way). Outside the marker map, the positions are specified explicitly in terms of recombination fractions with the nearest marker locus. Note that an external recombination fraction of 0.5 is not necessary since the likelihood of unlinked trait locus is always used as reference when computing the LOD scores.

The `set trait ... freqs' statement specifies allele frequencies at the trait locus. If the allele frequencies sum to below 1, a warning message will be issued:

Sum of allele frequencies is not in range .9999, 1.0001 (W)

Instead, if the allele frequencies sum to above 1.0001, the program quits and generates an error message.

The final five statements give MCMC specifications. The `sample by scan' statement asks that each iteration update all the meiosis indicators, S. The alternative `sample by step' updates one locus (L-sampler) or one meiosis (M-sampler) only in each iteration. In the `set burn-in iterations' statement, 100 burn-in iterations, with an unlinked trait, are requested. The L-sampler probability is set at 20 percent, which seems to be a good choice. For a detailed discussion of effects of varying L- to M-sampler ratio, see section 10.6 in Thompson (2000). The next statement requests 3000 MCMC iterations per test position of the trait locus. This is for demonstration purposes only. For real data analysis, use longer runs, on the order of 10^5 iterations per test position. With the last statement, the program checks the progress every 1000 iterations.

To specify different types of trait, different statements are needed. For a genotypic trait as in `ped73_ge.par', the part specifying the trait is

set trait data genotypic input pedigree record trait 1 integer 3

`ped73_ph.par' specifies a phenotpic trait as follows:

set trait data discrete input pedigree record trait 1 integer 3 set incomplete penetrance 0.05 0.6 0.95

And a quantitative trait is specified in `ped73_qu.par':

set trait data quantitative input pedigree record trait 1 real 1 set trait 1 genotype mean 90.0 100.0 110.0 set trait 1 residual variance 25.0

All three files use `input pedigree record trait' to specify which column in that file to be used as the trait value. The `set trait data' statement specifies the type of trait. Penetrances are needed for a phenotypic trait, whereas genotypic means and residual variance are needed for a quantitative trait. Additive variance may be needed to run lm_schnell and can be specified using the statement `set trait ... additive variance'. The default is 0.

Note that lm_markers can do the peeling by component using statement `set peeling by component'. Parameter file `ped73_qu_cmpt.par' has this option turned on, everything else being the same as file `ped73_qu.par'.

Since lm_bayes works for genotypic and discrete traits, parameters files `ped73_ge.par' and `ped73_ph.par' run with lm_bayes.

Parameter files for lm_lods and lm_schnell require slight modification: because MCMC is performed at each position, the number of MC iterations is much smaller in general than that used for lm_markers. In `ped73_lods_ph.par' and `ped73_schn_qu.par', the number of MC iterations is set to be 300 instead of 3000.