Cervus analyses genetic data from co-dominant genetic markers such as microsatellites (STRs) and SNPs. It assumes that the species is diploid and that markers are autosomal, although sex-linked markers can be used for some analyses. It also assumes that markers are inherited independently of each other, in other words that they are in linkage equilibrium.
Given genotypes that fulfil these assumptions Cervus can perform the following analyses:
Allele frequency analysis
Parentage testing using likelihood requires allele frequencies. Starting from a file of genotypes, this analysis calculates the frequency of each allele for each locus in the population, along with a range of summary statistics including tests for Hardy-Weinberg equilibrium and the presence of null alleles. These statistics help to determine the suitability of loci for downstream analysis.
Simulation of parentage analysis
The simulation of parentage analysis is useful for two reasons. First, simulation can be used to examine the feasibility of parentage analysis using a given set of loci. Second, simulation can be used to calculate critical values of likelihood ratios, so that when parentage analysis is carried out using real data, the confidence of parentage assignments can be determined.
Four different types of parentage tests can be simulated: maternity analysis, paternity analysis, parent pair analysis where the sexes of candidate parents are known and parent pair analysis where the sexes of candidate parents are unknown.
The simulation is based on the allele frequencies and the number of candidate parents to be tested for each offspring. The simulation also takes account of factors that may hinder analysis, including the presence of unsampled candidate parents and gaps and errors in the genotype data.
The parentage analysis module is designed to make the time-consuming task of testing many candidate parents against many offspring a relatively straightforward task, with clearly interpretable results. For each offspring tested, parentage is either assigned to the most-likely candidate parent with a pre-determined level of confidence, or is left unassigned.
Providing that no unreasonable assumptions are made in constructing lists of candidate parents, and that resolving power of the loci is sufficient, the output is suitable for unbiased estimation of individual reproductive success and its variance, as well as for pedigree reconstruction.
This analysis is used to identify individuals that have been resampled, even if the two genotypes do not match precisely. Identity analysis is particularly useful for studies using non-invasive tissue sampling or where samples are obtained from unmarked individuals.