PATRI (PaTeRnity Inference) is a program for paternity analysis of genetic data. The program requires genotypic, diploid data from one or more loci from mother-offspring pairs and from potential fathers. Typical data might include microsatellite markers, Restriction Fragment Length Polymorphisms (RFLPs) or Single Nucleotide Polymorphisms (SNPs). Given such genotypic data, PATRI can calculate posterior probabilities of paternity for all sampled offspring. When behavioral or ecological information can be used to divide the sampled males into different groups, PATRI can perform maximum likelihood analyses of hypotheses regarding the relative reproductive success of those groups. The underlying statistical methodology was described in Nielsen,R., Mattila,D.K., Clapham,P.J. and Palsbøll,P.J. 2001. Statistical Approaches to Paternity Analysis in Natural Populations and Applications to the North Atlantic Humpback Whale. Genetics 157:1673-1682.

    For all genotypes, PATRI can estimate the posterior probability that a particular male has sired a particular offspring, assuming a uniform prior among all males in the population. The male population size (N) can either be specified by the user as a fixed value, or uncertainty regarding N can be modeled using a uniform or Gaussian prior. Using a uniform prior corresponds to assuming no prior information regarding the male population size, except that an upper bound can be specified. PATRI can also produce a maximum likelihood estimate of N based solely on the parent-offspring genotypic data. The estimation of N assumes equal fecundity and unbiased sampling of males.

    If sampled males can be divided into groups based on behavioral or ecological information, PATRI can be used to evaluate hypotheses regarding the relative reproductive success of these groups. For k groups the user starts with a full model containing k-1 parameters, a2, a3,\u2026,ak, where ai, is defined as the reproductive success of group i relative to group 1. The user can then enter restrictions on these parameters. For example, the hypothesis that males from groups i and j have equal reproductive corresponds to the restriction a i = a j. Given a set of restrictions, PATRI can 1) maximize the likelihood and 2) plot a profile likelihood surface for any particular a i. The profile likelihood surface for ai is constructed by optimizing over all aj, j ¹ i. The maximum likelihood values are stored in a table, allowing the user to perform likelihood ratio tests of various hypotheses regarding reproductive success. This analysis can be done using a fixed value of N or by assuming N is uniformly or Gaussian distributed.

  • Windows Executable
  • Executable for Linux on Sun processor
  • Executable for Linux on Intel processor
  • Documentation (Readme file)
  • Example infile