Comparison of statistical methods for identification of Streptococcus thermophilus, Enterococcus faecalis, and Enterococcus faecium from randomly amplified polymorphic DNA patterns - PubMed (original) (raw)
Comparative Study
Comparison of statistical methods for identification of Streptococcus thermophilus, Enterococcus faecalis, and Enterococcus faecium from randomly amplified polymorphic DNA patterns
G Moschetti et al. Appl Environ Microbiol. 2001 May.
Abstract
Thermophilic streptococci play an important role in the manufacture of many European cheeses, and a rapid and reliable method for their identification is needed. Randomly amplified polymorphic DNA (RAPD) PCR (RAPD-PCR) with two different primers coupled to hierarchical cluster analysis has proven to be a powerful tool for the classification and typing of Streptococcus thermophilus, Enterococcus faecium, and Enterococcus faecalis (G. Moschetti, G. Blaiotta, M. Aponte, P. Catzeddu, F. Villani, P. Deiana, and S. Coppola, J. Appl. Microbiol. 85:25-36, 1998). In order to develop a fast and inexpensive method for the identification of thermophilic streptococci, RAPD-PCR patterns were generated with a single primer (XD9), and the results were analyzed using artificial neural networks (Multilayer Perceptron, Radial Basis Function network, and Bayesian network) and multivariate statistical techniques (cluster analysis, linear discriminant analysis, and classification trees). Cluster analysis allowed the identification of S. thermophilus but not of enterococci. A Bayesian network proved to be more effective than a Multilayer Perceptron or a Radial Basis Function network for the identification of S. thermophilus, E. faecium, and E. faecalis using simplified RAPD-PCR patterns (obtained by summing the bands in selected areas of the patterns). The Bayesian network also significantly outperformed two multivariate statistical techniques (linear discriminant analysis and classification trees) and proved to be less sensitive to the size of the training set and more robust in the response to patterns belonging to unknown species.
Figures
FIG. 1
(A) Schematic representation of an artificial neuron. The neuron is a simple processing unit connected to other neurons by synapses. A synaptic weight (wi) is associated with each synapsis. An output y is produced by using the weighted sum (z = Σxiwi) of its inputs (xi; x0 is fixed, and the product x0w0 is known as bias) as an argument of the activation function f(z). Different types of activation functions (nonlinear sigmoid functions as the logistic and hyperbolic tangent, but also threshold or linear functions) can be used. (B) Architecture of the ANNs used in this study. All types of networks used as an input the number of bands in selected molecular weight (in kilobases) intervals of the RAPD-PCR patterns and had three output nodes, one for each of the three species to be identified (EFM, E. faecium; EFS, E. faecalis; and ST, S. thermophilus). Both the MLP and the BN had a hidden layer with five nodes and used hyperbolic tangent activation functions, but they differed in the algorithm used to iteratively adjust the synaptic weights during supervised training (see the text for details). The hidden layer of the RBF was made up of 25 centers. For each of these, the Euclidean distance between an input pattern and the center was used as an argument of a nonlinear radial basis function, and the result was passed to the output nodes, which in turn had a linear activation function. The number and coordinates of the centers in the input space and the synaptic weights of the output neurons were adjusted during supervised training.
FIG. 2
Ethidium bromide-stained 1.5% (wt/vol) agarose gel displaying RAPD patterns of 32 strains of thermophilic streptococci obtained with primer XD9 (5′GAAGTCGTCC). Strain designations are shown above the lanes. Lane M, 1-kb DNA ladder (Gibco BRL) used as molecular size marker.
FIG. 3
Abridged dendrogram showing the similarity relationships among RAPD-PCR patterns of 138 strains of thermophilic streptococci. Percent similarity was calculated with the formula of Nei and Li (24), while clustering was carried out using UPGMA.
FIG. 4
Canonical score plot of simplified RAPD-PCR patterns obtained with primer XD9 for 138 strains of thermophilic streptococci. The canonical scores were calculated by discriminant analysis for the identification of S. thermophilus (○), E. faecalis (▵), and E. faecium (□) using RAPD-PCR patterns for a set of 93 strains (Table 1, group a). Other symbols: ♦, Streptococcus spp.; ▾, other enterococci. Open symbols correspond to patterns used for building the model; closed symbols correspond to patterns not used for building the model. The 95% confidence ellipses for the patterns of each species used for building the model are also shown.
FIG. 5
Dichotomic key generated by CT for the identification of S. thermophilus, E. faecalis, and E. faecium using RAPD-PCR patterns for a training set of 93 strains (Table 1, group a).
FIG. 6
Score plot for the principal-component analysis carried out on the outputs of a BN trained to identify S. thermophilus (○), E. faecalis (▵), and E. faecium (□) using RAPD-PCR patterns for a set of 93 strains (Table 1, group a). The output for all 138 strains of Table 1 is shown. Other symbols: ♦, Streptococcus spp.; ▾, other enterococci. Open symbols correspond to patterns used for building the model; closed symbols correspond to patterns not used for building the model. The 95% confidence ellipses for the patterns of each species used for building the model are also shown.
References
- Al-Haddad L, Morris C W, Boddy L. Training radial basis function neural networks: effects of training set size and imbalanced training sets. J Microbiol Methods. 2000;43:33–44. -PubMed
- Almeida J S, Sonesson A, Ringelberg D B, White D C. Application of artificial neural networks to the detection of Mycobacterium tuberculosis, its antibiotic resistance and prediction of pathogenicity amongst Mycobacterium spp. based on signature lipid biomarkers. Bin Comput Microbiol. 1995;7:159–166.
- Berthier F, Ehrlich S D. Genetic diversity of Lactobacillus sakei and Lactobacillus curvatus and design of PCR primers for its detection using randomly amplified polymorphic DNA. Int J Syst Bacteriol. 1999;49:997–1007. -PubMed
- Breiman L, Friedman J H, Olshen R A, Stone C I. Classification and regression trees. Belmont, Calif: Wadsworth; 1984.
Publication types
MeSH terms
LinkOut - more resources
Full Text Sources