Animal-side serologic assay for rapid detection of Mycobacterium Bovis infection in multiple species of free-ranging wildlife

Numerous species of mammals are susceptible to Mycobacterium bovis, the causative agent of bovine tuberculosis (TB). Several wildlife hosts have emerged as reservoirs of M. bovis infection for domestic livestock in different countries. In the present study, blood samples were collected from Eurasian badgers (n = 1532), white-tailed deer (n = 463), brushtail possums (n = 129), and wild boar (n = 177) for evaluation of antibody responses to M. bovis infection by a lateral-flow rapid test (RT) and multiantigen print immunoassay (MAPIA). Magnitude of the antibody responses and antigen recognition patterns varied among the animals as determined by MAPIA; however, MPB83 was the most commonly recognized antigen for each host studied. Other seroreactive antigens included ESAT-6, CFP10, and MPB70. The agreement of the RT with culture results varied from 74% for possums to 81% for badgers to 90% for wild boar to 97% for white-tailed deer. Small numbers of wild boar and deer exposed to M. avium infection or paratuberculosis, respectively, did not cross-react in the RT, supporting the high specificity of the assay. In deer, whole blood samples reacted similarly to corresponding serum specimens (97% concordance), demonstrating the potential for field application. As previously demonstrated for badgers and deer, antibody responses to M. bovis infection in wild boar were positively associated with advanced disease. Together, these findings suggest that a rapid TB assay such as the RT may provide a useful screening tool for certain wildlife species that may be implicated in the maintenance and transmission of M. bovis infection to domestic livestock.

An Extended Kernel for Generalized Multiple-Instance Learning

The multiple-instance learning (MIL) model has been successful in areas such as drug discovery and content-based image-retrieval. Recently, this model was generalized and a corresponding kernel was introduced to learn generalized MIL concepts with a support vector machine. While this kernel enjoyed empirical success, it has limitations in its representation. We extend this kernel by enriching its representation and empirically evaluate our new kernel on data from content-based image retrieval, biological sequence analysis, and drug discovery. We found that our new kernel generalized noticeably better than the old one in content-based image retrieval and biological sequence analysis and was slightly better or even with the old kernel in the other applications, showing that an SVM using this kernel does not overfit despite its richer representation.