CASPAR (Computerized Affected Sibling Pair Analyzer and Reporter)
PubMed Entrez BLAST OMIM Taxonomy Structure





CASPAR is an exploratory program to study the genetics of complex (polygenic) diseases. Given genetic information about families at multiple loci, it allows the user to explore hypotheses about how different factors may be involved in disease susceptibility of polygenic diseases. It semi-automatically and rapidly does an extended form of Affected Sibling Pair (ASP) tests for subset of families defined by their characteristics at one or more loci. The main advantage of CASPAR over othe ASP software packages is that CASPAR allows the user to do linkage analysis at one locus, conditional on the status of another locus, while properly accounting for the multiple statistical testing that is being done. Many linkage studies do such conditional linkage analysis, but ignore the fact that when the set of families is subdivided and a linkage is observed in one subset, one must account for the fact that the other subsets of families were also tested. By doing analysis with CASPAR, one can get some idea about how the genes in different linked regions may interact to cause disease susceptibility.



CASPAR carries out ASP analysis for multigene diseases and calculates sharing proportions and LOD scores. When run in simulation mode, it also computes p-values. The user chooses one or more loci suspected to be disease loci as the loci to condition on. These loci to condition on are called binning loci. The user specifies how to divide the families into (not necessarily disjoint) subsets based on the binning loci. Using CASPAR one looks for putative interactions between the binning loci and other loci which may interact with the binning loci.

CASPAR allows one to look at subsets of the data with certain properties. We achieve this by creating bins where a bin is defined by a condition. A sibpair qualifies for a bin if and only if it satisfies the bin condition. The binning methods supported at present are as follows:

  • Binning by genotype
  • Binning by sharing pattern
  • Binning by Identity by Descent (IBD) percentage
  • Binning by Identity by State (IBS)

We can create bins by looking at multiple loci at the same time. For example, if locus k is binned by genotype with the condition 1 2 x 2 3 and locus j is binned by sharing pattern with condition (share father's allele, not share mother's allele), then a sibpair qualifies for this bin if and only if one sib has alleles 1 2 at locus k, other sib has alleles 2 3 at locus k, and they have sharing pattern (share father's allele, not share mother's allele) at locus j. The reference for the paper describing CASPAR is:

  • Buhler J, Owerbach D, Schäffer AA, Kimmel M, Gabbay KH: Linkage Analyses in Type I Diabetes Using CASPAR, a Software and Statistical Program for Conditional Analysis of Polygenic Diseases. Human Heredity 47:211-222, 1997. [PubMed]

CASPAR as described in the above paper was for nuclear families containing at most two children and could not handle missing information. Since then, we have extended ASP test to handle families with any number of affected sibs, multiple generations, and parents whose genetic information is not known at some loci. We have also extended the multiple testing that CASPAR does by simulating given data in order to measure significance of the results to incorporate these features.

CASPAR can handle pedigrees with the following features.

  • Sibships of any size
  • Ungenotyped parents
  • Ungenotyped sibs: Sibs that are not genotyped at sharing locus or at a binning locus which is binned by any method other than by IBD percentage are ignored for analysis.
  • Unknown affection status: sibs with unknown affection status are simply ignored for analysis.

CASPAR handles ungenotyped parents by looping over all possible genotypes that yield compatible sibships and computing the likelihood of each choice using allele frequencies. The sharing result for each genotype choice gets appropriately weighted by the probability of that genotype being the true genotype.



The current software for CASPAR should not be used when one of the regions of interest is on a gender specific chromosome.


Send comments, questions, and suggestions to Richa Agarwala and Alejandro Schäffer




[Help] [Search]     [NLM NIH] [Disclaimer]