Next Previous Contents

9. The sib_kin program

The sib_kin program estimates the likelihood of an observed set of linkage data, for pairs of individuals, given different assumptions about their degree of relatedness. It generates likelihoods under the (generally false) assumption that all markers are independent, so the actual lod scores should be interpreted with caution. It can be used to verify family structures and/or check for some kinds of sample identification errors. It works by comparing the observed identity by state statistics for each pair with what would be expected by chance, given the observed allele frequencies.

The power of sib_kin is proportional to the number of informative and unlinked loci in the input data. For best results, the input data should consist of reasonably evenly spaced markers spanning multiple chromosomes. Typical genome scan data is ideal.

It should be noted that sib_kin impicitly assumes that the sample is in Hardy-Weinberg equilibrium, and wild deviations from this will lead to uninterpretable results. For example, data with alleles renumbered within families should not be used with this program.

9.1 Command syntax

sib_kin [-v] [-c cmd] [-f file] [marker_data ...] [> ibs_data]

One or more marker data files can be listed on the command line. If no files are specified, marker data will be read from standard input. Identity by state and likelihood data for relative pairs will be written to standard output.

9.2 TCL parameters

The following additional parameters can be specified using TCL commands via either the -c or -f mechanisms:


A boolean value: specifies that IBS and likelihood information should be reported for relative pairs with putative parents, as well as between sibs. The default is false.


A boolean value: specifies that a four-way comparison of the likelihoods of each pair being parent-sib, full sibs, half sibs, or unrelated should be performed. The default is false.

Here is a sample parameter file for sib_kin:

set nloc 100
set loc {
  D1S100 D1S200 D1S300
  D22S100 D22S200 D22S300
set blank "0"
set most_likely true

A sib_kin parameter file should either specify most_likely, or four_way, or give a specific set of z parameters to identify what relationship(s) are to be tested. Since sib_kin assumes that all markers are unlinked, the marker list may span multiple chromosomes, and markers need not be in any particular order. For best power, the markers should span all the autosomes.

9.3 Output

The output of sib_kin is a table with one line of output per relative pair, grouped by family. Non-verbose output leaves out families where all pairwise relationships are consistent with the family structure given in the data files; verbose output includes results for all families. Observed IBS statistics are reported, along with one or more log-likelihoods. The null hypothesis for each likelihood is that the pair is unrelated. If most_likely is true, then the maximum likelihood set of z values and corresponding LOD score are reported. If four_way is true, then log-likelihoods for a standard set of four relationships will be reported. Otherwise, the likelihoods for a pre-specified set of z parameters will be reported.

For the four-way analysis, in addition to the LOD scores for the four hypothetical relationships (where the score for the pair being unrelated is 0.0), sib_kin will identify which relationship was most likely, as either ``U'', ``I'', ``P'', ``S'', or ``H'', (for unrelated, identical, parent, sibling, or half-sib) if that relationship is at least 3 LOD units more likely than any of the other possibilities. If no relationship meets the cutoff, then ``?'' will be indicated.

Next Previous Contents