A simple 2D non-parametric resampling statistical approach to assess confidence in species identification in DNA barcoding--an alternative to likelihood and bayesian approaches

• Published in: PloS one Vol. 7; no. 12; p. e50831
• Main Authors: Jin, Qian, He, Li-Jun, Zhang, Ai-Bing
• Format: Journal Article
• Language: English
• Published: United States Public Library of Science 11.12.2012; Public Library of Science (PLoS)
• Subjects: Algorithms; Analysis; Animals; Bar codes; Bayes Theorem; Bayesian analysis; Biodiversity; Biology; Butterflies - genetics; Computer Simulation; Confidence; Datasets; Deoxyribonucleic acid; DNA; DNA Barcoding, Taxonomic; Evolution; Gene sequencing; Genetics, Population; Hybridization; Identification; Life sciences; Likelihood Functions; Likelihood ratio; Mathematical models; Mathematics; Methods; Mitochondrial DNA; Models, Genetic; Models, Theoretical; Nonparametric statistics; Nucleotide sequence; Phylogenetics; Population (statistical); Population genetics; Queries; Resampling; Simulation; Software; Species; Species Specificity; Statistics; Statistics, Nonparametric; Studies; Trees; Beijing China; China
• ISSN: 1932-6203; 1932-6203
• DOI: 10.1371/journal.pone.0050831

In the recent worldwide campaign for the global biodiversity inventory via DNA barcoding, a simple and easily used measure of confidence for assigning sequences to species in DNA barcoding has not been established so far, although the likelihood ratio test and the bayesian approach had been proposed to address this issue from a statistical point of view. The TDR (Two Dimensional non-parametric Resampling) measure newly proposed in this study offers users a simple and easy approach to evaluate the confidence of species membership in DNA barcoding projects. We assessed the validity and robustness of the TDR approach using datasets simulated under coalescent models, and an empirical dataset, and found that TDR measure is very robust in assessing species membership of DNA barcoding. In contrast to the likelihood ratio test and bayesian approach, the TDR method stands out due to simplicity in both concepts and calculations, with little in the way of restrictive population genetic assumptions. To implement this approach we have developed a computer program package (TDR1.0beta) freely available from ftp://202.204.209.200/education/video/TDR1.0beta.rar.