Hojin Moon

Highly Pathogenic Avian Influenza Virus (H5N1) in Domestic Poultry and Relationship with Migratory Birds, South Korea

During the 2006–2007 winter season in South Korea, several outbreaks of highly pathogenic avian influenza virus (H5N1) were confirmed among domestic poultry and in migratory bird habitats. Phylogenetic analysis showed that all isolates were closely related and that all belong to the A/bar-headed goose/Qinghai/5/2005–like lineage rather than the A/chicken/Korea/ES/2003–like lineage.

Classification by ensembles from random partitions of high-dimensional data

A robust classification procedure is developed based on ensembles of classifiers, with each classifier constructed from a different set of predictors determined by a random partition of the entire set of predictors. The proposed methods combine the results of multiple classifiers to achieve a substantially improved prediction compared to the optimal single classifier. This approach is designed specifically for high-dimensional data sets for which a classifier is sought. By combining classifiers built from each subspace of the predictors, the proposed methods achieve a computational advantage in tackling the growing problem of dimensionality. For each subspace of the predictors, we build a classification tree or logistic regression tree. Our study shows, using four real data sets from different areas, that our methods perform consistently well compared to widely used classification methods. For unbalanced data, our approach maintains the balance between sensitivity and specificity more adequately than many other classification methods considered in this study.

Rapid evolution of low-pathogenic H9N2 avian influenza viruses following poultry vaccination programmes.

To investigate whether currently circulating H9N2 avian influenza viruses (AIVs) in domestic poultry have evolved in Korean poultry since 2007, genetic and serological comparisons were conducted of H9N2 isolates from poultry slaughterhouses from January 2008 to December 2009. The isolation rate was relatively low in 2008 but increased gradually from January 2009 onwards. Genetic and phylogenetic analyses revealed that reassortant viruses had emerged, generating at least five novel genotypes, mostly containing segments of a previously prevalent domestic H9N2 virus lineage (Ck/Korea/04116/04-like). It was noteworthy that the N2 genes of some H9N2 isolates (genotypes D, E and F) were derived from those of H3N2-like viruses commonly isolated among domestic ducks in live-poultry markets. Animal challenge studies demonstrated that the pathogenicity of Ck/Korea/SH0906/09 (genotype B) and Ck/Korea/SH0912/09 (genotype F) in domestic avian species was altered due to reassortment. Furthermore, serological analysis revealed that the isolates were antigenically distinct from previous Korean H9N2 viruses including Ck/Korea/01310/01. Such antigenic diversity was illustrated further in experiments using H9N2-immunized chickens, which could not inhibit the replication and transmission of challenge viruses from each genotype. These results suggest that H9N2 viruses from domestic poultry have undergone substantial evolution since 2007 by immune selection as a result of vaccinal and natural immunity, coupled with reassortment. Taken together, this study demonstrates that periodical updating of vaccine strains, based on continuous surveillance data, is an important issue in order to provide sufficient protectivity against AIV infections.

Ensemble methods for classification of patients for personalized medicine with high-dimensional data

OBJECTIVE Personalized medicine is defined by the use of genomic signatures of patients in a target population for assignment of more effective therapies as well as better diagnosis and earlier interventions that might prevent or delay disease. An objective is to find a novel classification algorithm that can be used for prediction of response to therapy in order to help individualize clinical assignment of treatment. METHODS AND MATERIALS Classification algorithms are required to be highly accurate for optimal treatment on each patient. Typically, there are numerous genomic and clinical variables over a relatively small number of patients, which presents challenges for most traditional classification algorithms to avoid over-fitting the data. We developed a robust classification algorithm for high-dimensional data based on ensembles of classifiers built from the optimal number of random partitions of the feature space. The software is available on request from the authors. RESULTS The proposed algorithm is applied to genomic data sets on lymphoma patients and lung cancer patients to distinguish disease subtypes for optimal treatment and to genomic data on breast cancer patients to identify patients most likely to benefit from adjuvant chemotherapy after surgery. The performance of the proposed algorithm is consistently ranked highly compared to the other classification algorithms. CONCLUSION The statistical classification method for individualized treatment of diseases developed in this study is expected to play a critical role in developing safer and more effective therapies that replace one-size-fits-all drugs with treatments that focus on specific patient needs.

Human organ/tissue growth algorithms that include obese individuals and black/white population organ weight similarities from autopsy data.

Physiologically based pharmacokinetic (PBPK) models need the correct organ/tissue weights to match various total body weights in order to be applied to children and the obese individual. Baseline data from Reference Man for the growth of human organs (adrenals, brain, heart, kidneys, liver, lungs, pancreas, spleen, thymus, and thyroid) were augmented with autopsy data to extend the describing polynomials to include the morbidly obese individual (up to 250 kg). Additional literature data similarly extends the growth curves for blood volume, muscle, skin, and adipose tissue. Collectively these polynomials were used to calculate blood/organ/tissue weights for males and females from birth to 250 kg, which can be directly used to help parameterize PBPK models. In contrast to other black/white anthropomorphic measurements, the data demonstrated no observable or statistical difference in weights for any organ/tissue between individuals identified as black or white in the autopsy reports.

Persistence of Escherichia coli O157:H7 in experimentally infected swine

These experiments determined the ability of Escherichia coli O157:H7 to colonize and persist in pigs simultaneously inoculated with other pathogenic E. coli strains. Three-months-old pigs were inoculated with a mixture of five E. coli strains. The mixture included two Shiga toxigenic E. coli (STEC) O157:H7 strains, two enterotoxigenic E. coli (ETEC) strains and one enteropathogenic E. coli (EPEC) strain. A high dose mixture with all five strains at 10(10)CFU/animal (CFU: colony forming units) and a low dose mixture with the STEC strains at 10(7)CFU and the EPEC and ETEC strains remaining at 10(10)CFU were used. The STEC strains persisted in the alimentary tracts of some pigs at 2 months post-inoculation, following inoculation with both the high and low dose mixtures. When all strains were given at 10(10)CFU (high dose) the STEC strains persisted in greater numbers and in more pigs than did the other E. coli strains. The results demonstrated that persistent colonization (> or =2 months) by E. coli O157:H7 can occur in pigs. These findings were similar to those reported from sheep inoculated with the same mixture of E. coli strains. The results are consistent with reports suggesting that pigs have the potential to be reservoir hosts for STEC O157:H7.

Investigation of the biological indicator for vaccine efficacy against highly pathogenic avian influenza (HPAI) H5N1 virus challenge in mice and ferrets.

To investigate the biological indicator for vaccine efficacy against HPAI H5N1 virus challenge of varying clades, two inactivated whole-virus H5N1 vaccines containing the hemagglutinin (HA) and neuraminidase (NA) genes of either clade 2.2 A/EM/Korea/W149/06 (RgKoreaW149/06 x PR8) or clade 2.5 A/Ck/Korea/ES/03 (RgKoreaES223N/03XPR8) virus in the background of A/PR/8/34 (H1N1) were generated by reverse genetics. Administration of the vaccines (2-dose 1.77, 3.5, 7.5 or 15microg of HA) elicited high HI titers in a dose-dependent manner. Mice immunized with RgKoreaW149/06 x PR8 were completely protected from challenge against wild-type A/EM/Korea/W149/06 without clinical signs of infection. RgKoreaES223N/03XPR8 could not protect mice at 1.77microg while all immunized ferrets were completely protected. Two-dose (7.5microg) vaccinated mice (HI titer > or =320) and triple dose (7.5 microg) vaccinated ferrets with RgKoreaES223N/03xPR8 (HI titer > or =640) protected vaccine recipients from mortality, inhibited nasal virus shedding and limited influenza virus tropism. Thus, these vaccines provided cross-protectivity in both models. More importantly, these results collectively suggested a positive correlation between vaccine-induced HI titers and inhibition of virus shedding including block of viral proliferation in major organs against a heterologous HPAI H5N1 virus. Although developing technologies or methods that will enable the reduction of administration dose/frequency remains to be resolved, our study demonstrated a considerable biological marker (> or =640 HI titer) for full protection of the vaccinated hosts that could provide a preliminary basis for the assessment of complete immunization.

Mucosal immunization with recombinant influenza hemagglutinin protein and poly gamma-glutamate/chitosan nanoparticles induces protection against highly pathogenic influenza A virus.

Intranasal administration of recombinant influenza hemagglutinin (rHA) antigen or inactivated virus with nanoparticles (NPs) composed of poly-γ-glutamic acid (γ-PGA) and chitosan which are safe, natural materials, and able to target the mucosal membrane as a mucosal adjuvant, could induce a high degree of protective mucosal immunity in the respiratory tract. Intranasal immunization with mixture of rHA antigen or inactivated virus and γ-PGA/chitosan nanoparticles (PC NPs) induced not only a high anti-HA immunoglobulin A (IgA) response in lung and IgG response in serum, including anti-HA neutralizing antibodies, but also an influenza virus-specific cell-mediated immune response. Also, PC NPs could function as a potential mucosal adjuvant when it was compared with the well-known mucosal adjuvant, cholera toxin (CT). Intranasal administration of rHA antigen or inactivated virus with PC NPs protected mice against challenge with a lethal dose of the highly pathogenic influenza A H5N1 virus. These results suggested that application of PC NPs with a subunit antigen of influenza produced by prokaryotic expression system provides several solutions to the current problems of the influenza vaccines using inactivated influenza virus. Moreover, our finding about a sufficient function of PC NPs to elevate vaccine efficacy led us to consider that it can be useful in clinical applications as a potent mucosal adjuvant with safety.

Decision threshold adjustment in class prediction

Standard classification algorithms are generally designed to maximize the number of correct predictions (concordance). The criterion of maximizing the concordance may not be appropriate in certain applications. In practice, some applications may emphasize high sensitivity (e.g., clinical diagnostic tests) and others may emphasize high specificity (e.g., epidemiology screening studies). This paper considers effects of the decision threshold on sensitivity, specificity, and concordance for four classification methods: logistic regression, classification tree, Fishers linear discriminant analysis, and a weighted k-nearest neighbor. We investigated the use of decision threshold adjustment to improve performance of either sensitivity or specificity of a classifier under specific conditions. We conducted a Monte Carlo simulation showing that as the decision threshold increases, the sensitivity decreases and the specificity increases; but, the concordance values in an interval around the maximum concordance are similar. For specified sensitivity and specificity levels, an optimal decision threshold might be determined in an interval around the maximum concordance that meets the specified requirement. Three example data sets were analyzed for illustrations.

Classification methods for the development of genomic signatures from high-dimensional data

Personalized medicine is defined by the use of genomic signatures of patients to assign effective therapies. We present Classification by Ensembles from Random Partitions (CERP) for class prediction and apply CERP to genomic data on leukemia patients and to genomic data with several clinical variables on breast cancer patients. CERP performs consistently well compared to the other classification algorithms. The predictive accuracy can be improved by adding some relevant clinical/histopathological measurements to the genomic data.