No abstract

Background: The assignment of taxonomic attributions to DNA fragments recovered directly from the environment is a vital step in metagenomic data analysis. Assignments can be made using rank-specific classifiers, which assign reads to taxonomic labels from a predetermined level such as named species or strain, or rank-flexible classifiers, which choose an appropriate taxonomic rank for each sequence in a data set. The choice of rank typically depends on the optimal model for a given sequence and on the breadth of taxonomic groups seen in a set of close-to-optimal models. Homology-based (e.g., LCA) and composition-based (e.g., PhyloPythia, TACOA) rankflexible classifiers have been proposed, but there is at present no hybrid approach that utilizes both homology and composition.

Feral ungulates, such as pigs, are highly destructive to island ecosystems and are therefore often the target of eradication efforts. To succeed in eradication, however, managers must address a question made formidable by the great difficulty of detecting animals at very low levels of abundance: how will we know when elimination has been achieved? We developed and tested a framework to address this problem in a program to remove feral pigs from Santa Cruz Island, California. In an unprecedented timeframe for an island of this size, the program has progressed to a point at which pigs can no longer be detected. We describe seven key attributes of our approach, and how they have increased the likelihood that our inability to detect additional pigs indicates successful eradication, rather than the pigs having become better at escaping detection. This approach represents an important advance in the practice of eradication that can serve as a model for increasing the pace and scale of island restoration around the world.

Objectives To investigate the effect of breed as a risk factor associated with humeral condylar fracture in skeletally immature dogs in the UK. Materials and Methods Retrospective study of dogs under 12 months of age that were presented with humeral condylar fracture to three specialist referral centres between 2015 and 2018. Data retrieved from medical records included breed, age, gender, neuter status, affected limb, fracture configuration and aetiology of the fracture. Breed population percentages were compared with those recorded by the UK Kennel Club. Results Of the 115 dogs with 118 fractures, French bulldogs (41%) and English springer spaniels (15%) were overrepresented: humeral condylar fractures were more commonly diagnosed in French bulldogs (odds ratio = 5.86) and English springer spaniels (odds ratio = 5.66) compared with mixed‐breed dogs. Lateral condylar fractures occurred in 70% of cases, with medial condylar fractures and Y/T fractures accounting for 9% and 21%, respectively. Median age at the time of fracture was 4 months (range 2 to 10 months). Clinical Significance French bulldogs and English springer spaniels were identified as being at potentially increased risk of humeral condylar fracture in skeletally immature dogs.

Determining the taxonomic lineage of DNA sequences is an important step in metagenomic analysis. Short DNA fragments from next-generation sequencing projects and microbes that lack close relatives in reference sequenced genome databases pose significant problems to taxonomic attribution methods. Our new classification algorithm, RITA (Rapid Identification of Taxonomic Assignments), uses the agreement between composition and homology to accurately classify sequences as short as 50 nt in length by assigning them to different classification groups with varying degrees of confidence. RITA is much faster than the hybrid PhymmBL approach when comparable homology search algorithms are used, and achieves slightly better accuracy than PhymmBL on an artificial metagenome. RITA can also incorporate prior knowledge about taxonomic distributions to increase the accuracy of assignments in data sets with varying degrees of taxonomic novelty, and classified sequences with higher precision than the current best rank-flexible classifier. The accuracy on short reads can be increased by exploiting paired-end information, if available, which we demonstrate on a recently published bovine rumen data set. Finally, we develop a variant of RITA that incorporates accelerated homology search techniques, and generate predictions on a set of human gut metagenomes that were previously assigned to different ‘enterotypes’. RITA is freely available in Web server and standalone versions.