The LASSIE MPS panel: Predicting externally visible traits in dogs for forensic purposes.

Predicting the outward appearance of dogs via their DNA, also known as Canine DNA Phenotyping, is a young, emerging field of research in forensic genetics. The few previous studies published in this respect were restricted to the consecutive analysis of single DNA markers, a process that is time- and sample-consuming and therefore not a viable option for limited forensic specimens. Here, we report on the development and evaluation of a Massively Parallel Sequencing (MPS) based molecular genetic assay, the LASSIE MPS Panel. This panel aims to predict externally visible as well as skeletal traits, which include coat color, coat pattern, coat structure, tail morphology, skull shape, ear shape, eye color and body size from DNA using 44 genetic markers in a single molecular genetic assay. A biostatistical naïve Bayes classification approach was applied to identify the most informative marker combinations for predicting phenotypes. Overall, the predictive performance was characterized by a very high classification success for some of the trait categories, and high to moderate success for others. The performance of the developed predictive framework was further evaluated using blind samples from three randomly selected dog individuals, whose appearance was well predicted.

Towards Forensic DNA Phenotyping for Predicting Visible Traits in Dogs.

The popularity of dogs as human companions explains why these pets regularly come into focus in forensic cases such as bite attacks or accidents. Canine evidence, e.g., dog hairs, can also act as a link between the victim and suspect in a crime case due to the close contact between dogs and their owners. In line with human DNA identification, dog individualization from crime scene evidence is mainly based on the analysis of short tandem repeat (STR) markers. However, when the DNA profile does not match a reference, additional information regarding the appearance of the dog may provide substantial intelligence value. Key features of the dog's appearance, such as the body size and coat colour are well-recognizable and easy to describe even to non-dog experts, including most investigating officers and eyewitnesses. Therefore, it is reasonable to complement eyewitnesses' testimonies with externally visible traits predicted from associated canine DNA samples. Here, the feasibility and suitability of canine DNA phenotyping is explored from scratch in the form of a proof of concept study. To predict the overall appearance of an unknown dog from its DNA as accurately as possible, the following six traits were chosen: (1) coat colour, (2) coat pattern, (3) coat structure, (4) body size, (5) ear shape, and (6) tail length. A total of 21 genetic markers known for high predicting values for these traits were selected from previously published datasets, comprising 15 SNPs and six INDELS. Three of them belonged to SINE insertions. The experiments were designed in three phases. In the first two stages, the performance of the markers was tested on DNA samples from dogs with well-documented physical characteristics from different breeds. The final blind test, including dogs with initially withheld appearance information, showed that the majority of the selected markers allowed to develop composite sketches, providing a realistic impression of the tested dogs. We regard this study as the first attempt to evaluate the possibilities and limitations of forensic canine DNA phenotyping.

Forensic characterization and statistical considerations of the CaDNAP 13-STR panel in 1,184 domestic dogs from Germany, Austria, and Switzerland.

Crime scene samples originating from domestic dogs such as hair, blood, or saliva can be probative as possible transfer evidence in human crime and in dog attack cases. In the majority of such cases canine DNA identification using short tandem repeat (STR) analysis is the method of choice, which demands, among others, a systematic survey of allele frequency data in the relevant dog populations. A set of 13 highly polymorphic canine STR markers was used to analyze samples of 1,184 dogs (including 967 purebred dogs) from the so-called DACH countries (Germany, Austria, Switzerland). This CaDNAP 13-STR panel has previously been validated for canine identification in a forensic context. Here, we present robust estimates of allele frequencies, which are essential to assess the weight of the evidence by estimating the probability of a matching DNA profile within the dog population under question, e.g. in the form of a random match probability (RMP). The geographical provenance of the tested dogs showed a negligible influence on the observed genotype variation. Therefore, we combined the STR data from all three countries into a single dog population sample (DPS). In contrast, pronounced genetic differentiation between dog breeds was found by principal component analysis and sub-structure analysis with the STRUCTURE software. These findings entailed the need to account for the effects of DPS breed composition on allele frequency estimates. A possible strategy, which was favored here, relies on collecting a DPS that is guided by the breed composition of the relevant dog population. In total, dogs from 166 different breeds were included in our DPS, 64 of them including at least 5 individuals (n = 771 dogs). Sampling reflected the abundance of breeds in the DACH countries with the following being the most common ones: German Shepherds (population frequency: 14.3%), Dachshunds (5.9%), Labrador Retrievers (3.9%), and Golden Retrievers (3.2%). The pedigree listing of the purebred dogs in our DPS ranked German Shepherds (DPS frequency 8.5%) first, followed by Labrador Retrievers (3.9%), Golden Retrievers (3%), and Dachshunds (2.5%). RMP values based on overall allele frequencies and accounting for substructure using FST between breeds ranged between 10-13 and 10-14 and represent a conservative approach of RMP assessment.

Dog breed affiliation with a forensically validated canine STR set.

We tested a panel of 13 highly polymorphic canine short tandem repeat (STR) markers for dog breed assignment using 392 dog samples from the 23 most popular breeds in Austria, Germany, and Switzerland. This STR panel had originally been selected for canine identification. The dog breeds sampled in this study featured a population frequency ≥1% and accounted for nearly 57% of the entire pedigree dog population in these three countries. Breed selection was based on a survey comprising records for nearly 1.9 million purebred dogs belonging to more than 500 different breeds. To derive breed membership from STR genotypes, a range of algorithms were used. These methods included discriminant analysis of principal components (DAPC), STRUCTURE, GeneClass2, and the adegenet package for R. STRUCTURE analyses suggested 21 distinct genetic clusters. Differentiation between most breeds was clearly discernable. Fourteen of 23 breeds (61%) exhibited maximum mean cluster membership proportions of more than 0.70 with a highest value of 0.90 found for Cavalier King Charles Spaniels. Dogs of only 6 breeds (26%) failed to consistently show only one major cluster. The DAPC method yielded the best assignment results in all 23 declared breeds with 97.5% assignment success. The frequency-based assignment test also provided a high success rate of 87%. These results indicate the potential viability of dog breed prediction using a well-established and sensitive set of 13 canine STR markers intended for forensic routine use.