9.2 Sample lm_auto parameter file

lm_auto uses the parameter file `jv_rep_auto.par':

input pedigree file 'jv_rep.ped' input seed file '../sampler.seed' select all markers traits 1 map gender F markers distances 25.5 25.5 25.5 25.5 map gender M markers distances 11.2 45.8 11.2 45.8 map gender F trait 1 marker 2 distances 12.8 map gender M trait 1 marker 2 distances 5.8 set markers 1 2 3 4 freqs .2 .2 .4 .1 .06 .04 set markers 5 freqs .3 .2 .3 .1 set trait 1 freqs .95 .05 set marker data 5 333 1 3 1 3 1 3 1 3 1 3 331 3 4 3 4 3 4 3 4 3 4 334 2 3 2 3 2 3 2 3 2 3 431 3 4 3 4 3 4 3 4 3 4 531 3 3 3 3 3 3 3 3 3 3 343 1 3 1 3 1 3 1 3 1 3 341 3 5 3 5 3 5 0 0 3 3 344 4 6 4 6 4 6 2 4 2 4 441 3 4 3 4 0 0 3 4 3 4 541 3 3 3 3 3 3 3 3 3 3 set window patterns 0 4 set locus window 3 set component 1 proband gametes 531 1 531 0 331 0 333 1 set component 2 proband gametes 541 1 541 0 set L-sampler probability 0.2 set MC iterations 2000

The trait values are specified in the parameter file and are coded as `1', `3', `4' or `0', corresponding to trait locus genotypes of `1 1', `1 2' (or `2 1'), `2 2' or `missing', respectively. Since there is no `input pedigree record trait' statement in the example parameter file, the default behavior is implemented and so the trait value is listed after the names and gender in the pedigree file. The specified pedigree file, `jv_rep.ped', is a 30-member, two-component pedigree in which the final individuals (named 531 and 541) have trait value `4'. All other individuals in the file have trait value `0'. Because the trait type is not specified in the parameter file via a `set trait data' statement, the trait type is assumed to be genotypic. This means that the trait locus genotype can be inferred from the trait value, i.e. there are three distinct trait values, each corresponding to a distinct genotype at the trait locus.

The `map' statements specify the marker map and trait position in terms of genetic distances (centiMorgan). In this example there are five markers with gender-specific maps. The trait locus position is measured from the marker to its left. In this example, the trait locus for males is between markers 2 and 3 at a distance of 12.8 cM to the left of marker 2 (See See genedrop mapping model parameters. The `set markers' statements specify the number and frequency of alleles for each marker. In the example, the first four markers each have six alleles (labeled 1--6) with frequencies 0.2, 0.2, 0.4, 0.1, 0.06 and 0.04. The fifth marker has four alleles with frequencies 0.3, 0.2, 0.3 and 0.1. The trait locus has two alleles; alleles `1' and `2' have frequencies 0.95 and 0.05, respectively. The `select' statement is analogous to genedrop's `simulate' statement (see genedrop computing requests).

The `set marker data' statement specifies the number of markers to be five. Following the `set marker data' statement are genotype data for typed individuals. Alternatively, lm_auto can read genotype data from a separate file specified with an `input marker data file' statement.

The `set window patterns' and `set locus window' statements instruct lm_auto to compute the probabilities that the gametes named in the `set proband gametes' statement have a particular ibd pattern (also called state) accross several loci.

Recall that in ibddrop, one can compute the probability of two gametes being ibd or not. The values in the `IBD' column of the output indicate whether the gametes specified in the `set proband gametes' statement are ibd (indicated by a `1') or not (indicated by a `0'). With lm_auto, the user can specify ibd patterns of interest over two or more loci.

The `set locus window' statement specifies the number of loci to be examined simultaneously, in this case 3. This statement was discussed briefly in the ibddrop example,(See Running ibddrop example and sample output. The `set window patterns' statement indicates that we are interested in patterns `0' and/or `4', which correspond to ibd patterns `1 1 1 1' and `1 1 2 2', respectively. That is, in component 1, we are interested in the probability that all four of the gametes named in the `set proband gametes' statement are ibd across 3-locus windows or that the first and second gametes (maternal and paternal haplotypes of individual 531) are ibd and the third and fourth gametes (maternal haplotype of individual 531 and paternal haplotype of individual 333) are ibd, but these two pairs are not ibd with each other.

Recall the output of the ibddrop program generated when using the parameter file `ibd.par'. In the section of the program output headed `Probabilities of IBD patterns', each of the ibd patterns listed in the leftmost column is associated with a label in the right-most column.

Probabilities of IBD patterns Proband gamete set 1: 541 0 541 1 341 0 343 1 pattern marker-1 marker-2 trait-1 marker-3 marker-4 marker-5 label 1 1 1 1 .0287 .0298 .0310 .0273 .0287 .0298 0 1 1 1 2 .0290 .0275 .0292 .0282 .0302 .0305 1 1 1 2 1 .0132 .0135 .0138 .0140 .0139 .0132 3

The `set window patterns' statement in the parameter file for lm_auto expects one or more of these labels, which instruct it to calculate the probabilities of the associated pattern(s). This means that you must run ibddrop before using lm_auto to compute multi-locus probabilities.

The `set proband gametes' statement is the key statement for lm_auto. It specifies which haplotypes are to be scored with ibd probabilities. The syntax is as follows, where N1, N2, ... are individual ID's and K1, K2, ... indicate the haplotype as paternal (1) or maternal (0):

set [component M proband gametes N1 K1 N2 K2 ...

In the example, `531 1' refers to the paternal (1) haplotype of individual `531'. The first statement requests scoring both haplotypes of 531, the maternal (0) haplotype of 331, and the paternal (1) haplotype of 333. Note that as of MORGAN V2.9, the number of proband gametes is limited to 10. See ibddrop statements, for more discussion of the `set proband gametes' statement.

As with all of MORGAN's MCMC-based programs, the user can specify the desired number of MC iterations using the `set MC iterations' statement, the desired number of burn-in iterations using `set burn-in iterations', and the probability that the L-sampler is selected instead of the M-sampler using `set L-sampler probability'. In this example, 2000 sampling iterations are to be performed, using the L-sampler 20 percent of the time. These iterations are preceded by burn-in iterations. Because the number of burn-in iterations is not specified, lm_auto will use the default value of 10 percent of the number of main iterations. In practice, one would run the MCMC sampler much longer than 2000 iterations (on the order of 10^5).