Field Genetics

Cervus: Functions of Cervus

This page describes the functions of Cervus 3.0 and the output that it provides.

Allele frequency analysis

The main purpose of allele frequency analysis is to identify and count the alleles at each locus in a genotype file in order to estimate their frequency in the population. Cervus also calculates various summary statistics for each locus. These include the number of individuals typed, the number of heterozygotes and homozygotes, the number of alleles, observed and expected heterozygosity and polymorphic information content (PIC).

Cervus also calculates average exclusion probabilities for each locus - for a first parent, a second parent, a parent pair, identity and sib identity. Additionally Cervus can carry out a goodness-of-fit Hardy-Weinberg equilibrium test and estimate the frequency of any null allele that is present.

Simulation of parentage analysis

Simulation of parentage analysis is used to assess the power of a set of markers to assign parentage, and allows the confidence of real parentage assignments to be assessed. Four different types of parentage tests can be simulated: maternity analysis, paternity analysis, parent pair analysis with known sexes and parent pair analysis with unknown sexes.

The simulation takes allele frequencies either from a text file or directly from the output of the allele frequency module. These allele frequencies are used both to generate simulated genotypes and in the calculation of likelihood ratios.

The simulation also takes account of the number of candidate parents, the proportion of candidate parents that are sampled, the proportion of loci typed and the proportion of loci mistyped. It is also possible to simulate selfing and other forms of inbreeding, and to simulate candidate parents related to the true parents or offspring.

By default likelihood ratios are calculated using the more powerful revised likelihood equations of Kalinowski et al. (2007). However for comparison with previous studies it is also possible to use the original likelihood equations of Marshall et al. (1998).

Following common practice in parentage testing, Cervus displays the likelihood ratio in the form of a LOD score, the natural log of the likelihood ratio. Users may select the LOD score or a derivative of it, Delta, in order to evaluate the confidence of parentage assignments.

Users can select two levels of confidence for parentage assigned, relaxed and strict. By default relaxed confidence is 80% and strict confidence is 95%.

For each level of confidence the output of the simulation calculates critical values of LOD or Delta predicts the success of parentage testing using these critical values - the proportion of parentage tests expected to be resolved with this level of confidence. Optionally users can apply the critical values from one simulation to a second simulation run with a different set of parameters.

Cervus also outputs a summary of the distributions of LOD or Delta scores and provides a breakdown of parentage assignments separately for true parents and non-parents. Optionally Cervus will output the distributions of LOD or Delta, histograms of these distributions and tables of confidence levels calculated from them. The tables can be used to determine a confidence value for each parentage assignment.

Parentage analysis

In parentage analysis candidate parents or parent pairs are assigned to offspring. Cervus does the real parentage analysis in exactly the same way as the simulated parentage analysis, except using real rather than simulated genotypes.

To carry out parentage analysis, Cervus needs to know the names of the offspring to be tested, the names of known parents (if available) and the names of the candidate parents to be tested against each offspring. The Parentage Wizard guides users through the process of setting up parentage analysis.

For each offspring Cervus calculates the likelihood of parentage of every candidate parent or parent pair. Confidence in assignment of parentage to the most likely candidate parent (or parent pair) is evaluated relative to the critical LOD or Delta scores determined in the simulation.

The output is summarised in the same way as the simulation, comparing the observed success of parentage assignment against the success rate predicted by the simulation. The summary also reports any problems encountered during analysis, for example it will list the IDs of individuals with missing genotypes. For paternity analysis with known mothers or maternity analysis with known fathers, Cervus uses the frequency of known parent-offspring mismatches to estimate the genotyping error rate.

The parentage data file can most conveniently be viewed using Microsoft Excel or similar software. For each offspring the data file shows the identity, LOD score and confidence level of the most likely candidate parent (or parent pair), and if the user requests it, the file also shows the identity and LOD score of other candidate parents (or parent pairs). Optionally the output can include the non-exclusion probability for each offspring and, in the case of maternity and paternity analysis, for the offspring-known parent pair.

Identity analysis

Identity analysis is used to find matching genotypes in a genotype file.

The identity analysis reads a text-based file of genotypes, and compares each row of genotype data against every other row in the file. Cervus records any IDs that occur more than once in the file, and separately records any genotypes that occur more than once.

Cervus offers a "fuzzy matching" function, so users can look for genotypes that match at all except one locus, for example, as well as exact matches. It is also possible to search for matches taking account of the sex of individuals while testing individuals of unknown sex against both sexes.

The identity data file can most conveniently be viewed using Microsoft Excel or similar software. The file shows the type of match found and calculates the probability of identity and the probability of sib identity for any exact matches.

File conversion

Cervus can convert genotype files to or from Cervus format. Three other formats are supported: the population genetics programs Genepop and Genetix, and the relatedness program Kinship. Genepop format can be used by a number of other programs besides Genepop itself; some other programs support formats similar to the Cervus genotype file format.

Cervus can also convert text files between Windows, Mac and Unix/Linux operating systems.

Further information

How to use these features and how to interpret the results are described in more detail in the help file provided with Cervus. Part of the help file about the methodology used by Cervus is reproduced here.