Microsatellite null alleles in parentage analysis

Heredity (2004) 93, 504509

& 2004 Nature Publishing Group All rights reserved 0018-067X/04 $30.00www.nature.com/hdyMicrosatellite null alleles in parentage analysisEE Dakin and JC AviseDepartment of Genetics, University of Georgia, Athens, GA 30602, USAalleles might impact molecular parentage analyses. The

results indicate that whereas null alleles in frequencies

typically reported in the literature introduce rather inconsequential biases on average exclusion probabilities, they

can introduce substantial errors into empirical assessments

of specific mating events by leading to high frequencies of

false parentage exclusions.Heredity (2004) 93, 504509. doi:10.1038/sj.hdy.6800545Published online 4 August 2004Keywords: exclusion probability; parentage; population genetics; nonamplifying allelesHighly polymorphic microsatellite markers are widely employed in population genetic analyses (eg, of biological

parentage and mating systems), but one potential drawback

is the presence of null alleles that fail to amplify to detected

levels in the PCR assays. Here we examine 233 published

articles in which authors reported the suspected presence of

one or more microsatellite null alleles, and we review how

these purported nulls were detected and handled in the data

analyses. We also employ computer simulations and

analytical treatments to determine how microsatellite nullIntroductionBackground and literature reviewNull alleles have posed recurring challenges for population genetics following the introduction and application

of each new method of molecular assay (including

serological typing, protein electrophoresis, RFLPs, and

now microsatellites). One of the earliest examples

involved the human ABO blood group system, wherein

the O allele is a null allele that produces no phenotype (ie

is masked by the presence of the A or B alleles, which are

codominant to each other). For the special ABO case,

formulae have been developed for estimating the

frequency of the O allele (Yasuda and Kimura, 1968) as

well as for calculating the average paternity exclusion

probability (Weir, 1996). In general, however, more

emphasis has been placed on finding codominant

markers than on improving statistical methods for

dealing with null alleles.By definition, a microsatellite null allele is any allele at

a microsatellite locus that consistently fails to amplify to

detected levels via the polymerase chain reaction (PCR).

Despite the frequent invocation of null alleles to explain

unexpected genotypic patterns in microsatellite data sets,

little has been reported regarding their impact on

statistical parameters used to draw biological inferences.

Here we review scientific articles that describe methods