National Cheng Kung University

About: National Cheng Kung University is a education organization based out in Tainan City, Taiwan. It is known for research contribution in the topics: Population & Thin film. The organization has 49723 authors who have published 69799 publications receiving 1437420 citations. The organization is also known as: NCKU.

Reliable Accuracy Estimates from k -Fold Cross Validation

Abstract: It is popular to evaluate the performance of classification algorithms by k -fold cross validation. A reliable accuracy estimate will have a relatively small variance, and several studies therefore suggested to repeatedly perform k -fold cross validation. Most of them did not consider the correlation among the replications of k -fold cross validation, and hence the variance could be underestimated. The purpose of this study is to explore whether k -fold cross validation should be repeatedly performed for obtaining reliable accuracy estimates. The dependency relationships between the predictions of the same instance in two replications of k -fold cross validation are first analyzed for k -nearest neighbors with $k= 1$ k = 1 . Then, statistical methods are proposed to test the strength of the dependency level between the accuracy estimates resulting from two replications of k -fold cross validation. The experimental results on 20 data sets show that the accuracy estimates obtained from various replications of k -fold cross validation are generally highly correlated, and the correlation will be higher as the number of folds increases. The k -fold cross validation with a large number of folds and a small number of replications should be adopted for performance evaluation of classification algorithms.

High-level variability in the ORF-K1 membrane protein gene at the left end of the Kaposi's sarcoma-associated herpesvirus genome defines four major virus subtypes and multiple variants or clades in different human populations.

Abstract: Infection with Kaposi's sarcoma (KS)-associated herpesvirus (KSHV) or human herpesvirus 8 (HHV8) is common in certain parts of Africa, the Middle East, and the Mediterranean, but is rare elsewhere, except in AIDS patients. Nevertheless, HHV8 DNA is found consistently in nearly all classical, endemic, transplant and AIDS-associated KS lesions as well as in some rare AIDS-associated lymphomas. The concept that HHV8 genomes fall into several distinct subgroups has been confirmed and refined by PCR DNA sequence analysis of the ORF-K1 gene encoding a highly variable glycoprotein related to the immunoglobulin receptor family that maps at the extreme left-hand end of the HHV-8 genome. Among more than 60 different tumor samples from the United States, central Africa, Saudi Arabia, Taiwan, and New Zealand, amino acid substitutions were found at a total of 62% of the 289 amino acid positions. These variations defined four major subtypes and 13 distinct variants or clades similar to those found for the HIV ENV protein. The B and D subtype ORF-K1 proteins differ from the A and C subtypes by 30 and 24%, respectively, whereas A and C differ from each other by 15%. In all cases tested, multiple samples from the same patient were identical. Examples of the B subtype were found almost exclusively in KS patients from Africa or of African heritage, whereas the rare D subtypes were found only in KS patients of Pacific Island heritage. In contrast, C subtypes were found predominantly in classic KS and in iatrogenic and AIDS KS in the Middle East and Asia, whereas U.S. AIDS KS samples were primarily A1, A4, and C3 variants. We conclude that this unusually high diversity, in which 85% of the nucleotide changes lead to amino acid changes, reflects some unknown powerful biological selection process that has been acting preferentially on this early lytic cycle membrane signalling protein. Two distinct levels of ORF-K1 variability are recognizable. Subtype-specific variability indicative of long-term evolutionary divergence is both spread throughout the protein as well as concentrated within two 40-amino-acid extracellular domain variable regions (VR1 and VR2), whereas intratypic variability localizes predominantly within a single 25-amino-acid hypervariable Cys bridge loop and apparently represents much more recent changes that have occurred even within specific clades. In contrast, numerous extracellular domain glycosylation sites and Cys bridge residues as well as the ITAM motif in the cytoplasmic domain are fully conserved. Overall, we suggest that rather than being a newly acquired human pathogen, HHV8 is an ancient human virus that is preferentially transmitted in a familial fashion and is difficult to transmit horizontally in the absence of immunosuppression. The division into the four major HHV8 subgroups is probably the result of isolation and founder effects associated with the history of migration of modern human populations out of Africa over the past 35,000 to 60,000 years.