Assessment of statistical methods used in library-based approaches to microbial source tracking

Abstract

Several commonly used statistical methods for fingerprint identification in microbial source tracking (MST) were examined to assess the effectiveness of pattern-matching algorithms to correctly identify sources. Although numerous statistical methods have been employed for source identification, no widespread consensus exists as to which is most appropriate. A large-scale comparison of several MST methods, using identical fecal sources, presented a unique opportunity to assess the utility of several popular statistical methods. These included discriminant analysis, nearest neighbour analysis, maximum similarity and average similarity, along with several measures of distance or similarity. Threshold criteria for excluding uncertain or poorly matched isolates from final analysis were also examined for their ability to reduce false positives and increase prediction success. Six independent libraries used in the study were constructed from indicator bacteria isolated from fecal materials of humans, seagulls, cows and dogs. Three of these libraries were constructed using the rep-PCR technique and three relied on antibiotic resistance analysis (ARA). Five of the libraries were constructed using Escherichia coli and one using Enterococcus spp. (ARA). Overall, the outcome of this study suggests a high degree of variability across statistical methods. Despite large differences in correct classification rates among the statistical methods, no single statistical approach emerged as superior. Thresholds failed to consistently increase rates of correct classification and improvement was often associated with substantial effective sample size reduction. Recommendations are provided to aid in selecting appropriate analyses for these types of data.

Identifying Host Sources of Fecal Pollution: Diversity of Escherichia coli in Confined Dairy and Swine Production Systems

“…This finding is in agreement with the large E. coli population diversity now having been characterized in other potential sources of fecal pollution (7,8,13,18). In any MST study, the choice of source reference library size will be determined by the size and complexity of the study area, the accuracy of source identity required by the investigators, and the financial resources available (13,19,21,27).…”

“…Typically with this approach, environmental isolates are compared with reference collections of bacteria obtained from potential sources of fecal pollution in the area and on the basis of similarity are ascribed a probable host source. For any given study, the likelihood of correct classification will depend on several factors, including the size and representability of the reference collection, geographic size of the study area, choice of methods for image analysis and pattern recognition, and statistical methods for comparison of environmental isolates with the reference collection (1,8,13,(19)(20)(21)(22)27). The construction of robust fingerprint libraries is expensive and therefore requires an informed sampling strategy.…”

High-Throughput and Quantitative Procedure for Determining Sources of Escherichia coli in Waterways by Using Host-Specific DNA Marker Genes

“…This is similar to macroarray analysis used to identify genomic clones containing specific DNA sequences (27). The inclusion of multiple positive and negative control organisms and replications of unknowns on the membrane increase the statistical power of the results (22) and reduce the number of false-positive and -negative results that are common to many MST methods (29).…”

Microbial Source Tracking in a Watershed Dominated by Swine

“…The host-origin library was developed and clonal isolates (duplicate ARPs) were identified and removed. Classification ties were assigned a source depending on where the isolate was observed in dendrograms [41]. Additional efforts to develop a stringent host origin library and to obtain reliable source identification of unknown source isolates involved the application of an 80% threshold criterion for correct classification to the library.…”

“…The average frequency of misclassification (AFM) was calculated by adding the percentage of isolates incorrectly classified from each source category and dividing by the total number of source categories [22,41].…”