Numerical classification and identification of Aeromonas genospecies.

A total of 176 Aeromonas strains representing all currently characterized genospecies were tested for 329 biochemical characters. Overall similarities of all strains were determined by numerical taxonomic techniques, the UPGMA algorithm and the SSM and the SJ coefficients as measures of similarity. Sixteen clusters (two or more strains) and seven unclustered strains were recovered at the 93.5% similarity level (SSM). Genospecies 1, 4, 5, 6, 7, 9, 12 and 13 were largely represented by single phena, whereas strains of genospecies 2 and 3 were found in closely-related phena. Strains belonging to genospecies 8 formed two distinct biotypes. Strains belonging to genospecies 11 formed a subcluster within a cluster representing different genospecies. In general, similar groupings were obtained with the Jaccard coefficient at a similarity level of 80.0% (SJ) with minor changes in the definition of clusters. The phenetic data showed good correlation with the taxa defined by DNA/DNA hybridization and those obtained by multilocus enzyme analysis. For all genospecies (independent from cluster assignment) 30 diagnostic characters were selected to construct a matrix for probabilistic identification. The correct identification rate of the matrix was 71.51% taking a Willcox probability greater than 0.99, and 83.7% taking a Willcox probability greater than 0.9 as identification threshold levels.     

A probability matrix for the identification of vibrios

A probability matrix for computer-assisted identification of vibrios has been constructed, based on the API 20E system. Data were gathered from 173 strains representing 31 taxa of vibrios and related organisms, from a variety of sources. The matrix was tested internally by four statistical programs. Program OVERMAT tested the separation and program MOSTTYP the discretion and homogeneity of the taxa. Most of the taxa were satisfactory but a few were less so; reasons for this are discussed. Program CHARSEP and program DIACHAR tested the separation and diagnostic values, respectively, of the characters used. The overall test error was 4.5%. The matrix was assessed externally by its performance in the identification of vibrio-like strains isolated from freshwater. Of 243 wild strains, 79.4% were identified with ten taxa, with a Willcox score of greater than or equal to 0.99.     

A probability matrix for the identification of vibrios

A probability matrix for computer-assisted identification of vibrios has been constructed, based on the API 20E system. Data were gathered from 173 strains representing 31 taxa of vibrios and related organisms, from a variety of sources. The matrix was tested internally by four statistical programs. Program OVERMAT tested the separation and program MOSTTYP the discretion and homogeneity of the taxa. Most of the taxa were satisfactory but a few were less so; reasons for this are discussed. Program CHARSEP and program DIACHAR tested the separation and diagnostic values, respectively, of the characters used. The overall test error was 4–5%. The matrix was assessed externally by its performance in the identification of vibrio-like strains isolated from freshwater. Of 243 wild strains, 79–4% were identified with ten taxa, with a Willcox score of ζ 0–99.     

Numerical classification and identification of Aeromonas genospecies

P. KÄMPFER AND M. ALTWEGG. 1992. A total of 176 Aeromonas strains representing all currently characterized genospecies were tested for 329 biochemical characters. Overall similarities of all strains were determined by numerical taxonomic techniques, the UPGMA algorithm and the S _SM and the S _J coefficients as measures of similarity. Sixteen clusters (two or more strains) and seven unclustered strains were recovered at the 93.5% similarity level ( S _SM). Genospecies 1, 4, 5, 6, 7, 9, 12 and 13 were largely represented by single phena, whereas strains of genospecies 2 and 3 were found in closely-related phena. Strains belonging to genospecies 8 formed two distinct biotypes. Strains belonging to genospecies 11 formed a subcluster within a cluster representing different genospecies. In general, similar groupings were obtained with the Jaccard coefficient at a similarity level of 80.0% ( S _J) with minor changes in the definition of clusters. The phenetic data showed good correlation with the taxa defined by DNA/DNA hybridization and those obtained by multilocus enzyme analysis. For all genospecies (independent from cluster assignment) 30 diagnostic characters were selected to construct a matrix for probabilistic identification. The correct identification rate of the matrix was 71.51% taking a Willcox probability < 0.99, and 83.7% taking a Willcox probability > 0.9 as identification threshold levels.     

Numerical classification and identification of Bacillus sphaericus including some strains pathogenic for mosquito larvae

Ninety-one strains of Bacillus sphaericus, including representatives of all the established DNA homology groups, related round-spored and oval-spored species, and six strains pathogenic for mosquito larvae, were examined for 155 characters. Numerical analyses (Jaccard coefficient/average linkage clustering) based on the 88 variable features revealed 14 clusters at the 79% similarity level that contained more than one strain and 17 single member clusters. All insect pathogenic strains were recovered in a single cluster and the classification was in accord with an established classification based on DNA sequence homology. Two frequency matrices for probabilistic identification were constructed and tested. A comprehensive matrix comprising 14 mesophilic, round-spored taxa and 27 tests gave good results for identification of hypothetical median organisms, cluster overlap and identifications of representative strains (based on data generated in the classification study). Reference strains for the 14 taxa and eight additional insect pathogenic strains were examined for the 27 tests and were correctly identified with high scores using this matrix. A second matrix comprising seven taxa and 13 tests also performed well in the theoretical evaluation and correctly identified the reference strains and insect pathogenic strains.     

Programs for evaluating and characterising bacterial taxonomic data

Three programs are described for evaluating and characterisingdata collected during numerical taxonomic studies of bacteria.The program VARIANCE compares replicate cultures and evaluatesthe reproducibility of each character. Also it identifies thosecharacters that should be excluded from subsequent taxonomicanalysis because of their poor reproducibility. GPROPS summarisesthe properties of clusters of strains that have been definedfrom a cluster analysis, it can produce a probabilistic identificationmatrix and compares each strain within a cluster with the HypotheticalMean Organism (HMO) of that cluster. OVCLUST is an implementationof the program described by Sneath (1979) which calculates overlapstatistics between major clusters. These programs are designedto complement the CLUSTAN package (Wishart, 1982) which is oftenused for cluster analysis of bacterial taxonomic data. The programswere written in FORTRAN 77 and implemented on an IBM PC usingMS–DOS. Received on November 13, 1986; accepted on January 8, 1987     

Numerical taxonomy and laboratory identification of Actinomyces and Arachnia and some related bacteria.

A numerical taxonomic study was made on 49 facultative anaerobic Gram-positive filamentous and/or diphtheroidal organisms isolated from dental plaques, carious dentin and faeces, together with 63 reference strains belonging to the genera Actinomyces, Arachnia, Bifidobacterium, Actinobacterium, Propionibacterium, Eubacterium and Lactobacillus. They were examined for 90 unit characters covering a wide range of tests and properties. The data were subjected to computer analysis in which the simple matching coefficient (SSM) and the similarity index (SJ) were calculated, and the results of single linkage techniques and an unweighted average linkage cluster analysis technique were compared. The strains fell into six major groups (phena). The Actinomyces strains were recovered in two phena; the first contained Actinomyces israelii and the other facultative anaerobic Actinomyces, including subclusters equal to taxospecies of A. odontolyticus and A. viscosus/A. naeslundii, while the other phenon corresponded to the genera Arachnia, Actinobacterium, Bifidobacterium and Propionibacterium. The groups of Arachnia and Actinobacterium each contained one species, representing taxospecies of Arachnia propionica and Actinobacterium meyerii. Taxonomic criteria, both constant and discriminative, were selected to form a diagnostic table useful for laboratory identification of this group of organisms. Immunofluorescence supported the numerical data.     

Software for the development and evaluation of probabilistic identification matrices

A number of programs are described for the development and evaluationof probabilistic identification matrices for use with computer-assistedidentification. The program BEST reads an initial matrix ofper cent probabilities for all the binary characters examinedduring a cluster analysis and determines the most useful setof tests for distinguishing taxa in the matrix. Program RESORTcreates from the initial matrix an identification matrix inwhich the order of the tests may be different or the numberof tests reduced. A printed version of a matrix can be producedby MATPRJNT, which creates tables giving the per cent probabilityof a positive test result and the test results presented as‘–’, ‘v’ and ‘+’ dependingon a user-specified threshold. Program IDSC evaluates an identificationmatrix by calculating the best identification score for eachtaxon in the matrix using the expected result for each test.The programs were written in FORTRAN 77 and can be run on anymicrocomputer using the PC/MS-DOS operating system. Received on April 27, 1990; accepted on November 5, 1990     

Three-dimensional cluster analysis identifies interfaces and functional residue clusters in proteins

Three-dimensional cluster analysis offers a method for the prediction of functional residue clusters in proteins. This method requires a representative structure and a multiple sequence alignment as input data. Individual residues are represented in terms of regional alignments that reflect both their structural environment and their evolutionary variation, as defined by the alignment of homologous sequences. From the overall (global) and the residue-specific (regional) alignments, we calculate the global and regional similarity matrices, containing scores for all pairwise sequence comparisons in the respective alignments. Comparing the matrices yields two scores for each residue. The regional conservation score (C(R)(x)) defines the conservation of each residue x and its neighbors in 3D space relative to the protein as a whole. The similarity deviation score (S(x)) detects residue clusters with sequence similarities that deviate from the similarities suggested by the full-length sequences. We evaluated 3D cluster analysis on a set of 35 families of proteins with available cocrystal structures, showing small ligand interfaces, nucleic acid interfaces and two types of protein-protein interfaces (transient and stable). We present two examples in detail: fructose-1,6-bisphosphate aldolase and the mitogen-activated protein kinase ERK2. We found that the regional conservation score (C(R)(x)) identifies functional residue clusters better than a scoring scheme that does not take 3D information into account. C(R)(x) is particularly useful for the prediction of poorly conserved, transient protein-protein interfaces. Many of the proteins studied contained residue clusters with elevated similarity deviation scores. These residue clusters correlate with specificity-conferring regions: 3D cluster analysis therefore represents an easily applied method for the prediction of functionally relevant spatial clusters of residues in proteins.     

Identification of some industrially importantActinoplanes species

Summary Three actinomycetes designated A294, A385 and A448 all produced novel inhibitors of an enzyme involved in the regulation of mammalian cholesterol metabolism. The organisms were identified as members of the genusActinoplanes using morphological and chemotaxonomic characters. Numerical phenetic data obtained for clusters defined at the 77.5% S_sm similarity level in the taxonomic study ofActinoplanes and related taxe by Goodfellow et al. [6], was used to construct a probabilistic identification matrix for the genusActinoplanes. This was in an attempt to characterise the producing strains to species level. Identification scores for the three organisms, together with two type strains, were determined using the MATIDEN program of Sneath [36]. Strains A294 and A385 did not identify correctly with any cluster. Strain A448 was identified asActinoplanes derwentensis. The two type strains identified within acceptable statistical limits to their correct clusters. Due to the absence of other taxonomic criteria with which to speciate members of the genusActinoplanes we recommend that the two unidentified producing strains should be considered novel until such time as surther techniques become available.     

Genetic variability and relationships among thirty genotypes of finger millet (Eleusine coracana L. Gaertn.) using RAPD markers

Ragi or finger millet (Eleusine coracana L.) is an important crop used for food, forage, and industrial products. It is distributed in tropical and temperate regions of the world. The germplasm identification and characterization is an important link between the conservation and utilization of plant genetic resources. Traditionally, species or varieties identification has relied on morphological characters like growth habit, leaf architecture or floral morphology. Investigation through RAPD (random amplified polymorphic DNA) markers was undertaken for identification and determination of the genetic variation among thirty genotypes of ragi of the family Poaceae. Thirteen selected decamer primers were used for genetic analysis. A total of 124 distinct DNA fragments ranging from 300-3000 bp was amplified by using selected random RAPD marker. The genetic similarity was evaluated on the basis of the presence or absence of bands. Cluster analysis was made by the similarity coefficient. It indicated that the 30 genotypes of ragi form two major clusters, first, a major cluster having only one genotype, i. e. Dibyasinha and a second major cluster having twenty-nine genotypes. The second major cluster again subdivides into two minor clusters. A first minor cluster has only three varieties, i. e. Neelachal, OEB-56 and Chilika. The genotypes Neelachal and OEB-56 exhibit a 86% similarity with each other and 80% similarity with Chilika. A second minor cluster has 26 genotypes and is divided into two sub-minor clusters. The first sub-minor cluster has only one genotype (VL-322). The second sub-minor cluster again subdivides into two groups. One group has one genotype and the second group again is divided into two sub-groups, one with 13 genotypes and the other with 11 genotypes. The highest similarity coefficient was detected in a genotype collected from southern India and the least from northern India. The genotypes of finger millet collected from diverse agroclimatic regions of India constitute a wide genetic base. This is helpful in breeding programs and a major input into conservation biology of cereal crop.     

应用16S rRNA基因V—2高变区序列进行链霉莲分子分类

用16S rRNA部分序列对Williams数值分类系统所包含的链霉菌属种或种群进行系统进货分析,以包含16S rRNA基因V－2高变区在内的120bp长核苷酸序列所作的系统进化树表明：这些链霉菌可分为34个簇,其中大簇5个（包括27－85株菌株）;中等大小簇3个（包括9－12株菌）;小簇8个（包括2－6株菌）;单成员簇19个。结果同时表明,Williams数值分析系统中根据形态、生理、生化特征进行归类而得到的种或种群内存在较大异质性。     

A probability matrix for the identification of species of Vibrio and related genera

A matrix for the probabilistic identification of species of Vibrio and related genera has been constructed using the data from 1091 strains collected throughout the world and classified. Thirty-eight phenons are included in the matrix, 31 of these represent previously identified species or biovars and seven represent phenons which could not be identified and may represent new species. The identification matrix incorporates 81 characters although a subset of 30 tests can be used to distinguish the 38 phenons from each other. The additional 51 tests were included to assist the identification of some strains for which the initial 30 tests were inadequate. No significant cluster overlap was found at the 5% level and the identification score for the Hypothetical Median Organism of each cluster exceeded 0.9999 in all cases.     

A probability matrix for the identification of species of Vibrio and related genera

A matrix for the probabilistic identification of species of Vibrio and related genera has been constructed using the data from 1091 strains collected throughout the world and classified. Thirty-eight phenons are included in the matrix, 31 of these represent previously identified species or biovars and seven represent phenons which could not be identified and may represent new species. The identification matrix incorporates 81 characters although a subset of 30 tests can be used to distinguish the 38 phenons from each other. The additional 51 tests were included to assist the identification of some strains for which the initial 30 tests were inadequate. No significant cluster overlap was found at the 5% level and the identification score for the Hypothetical Median Organism of each cluster exceeded 0.9999 in all cases.     

Identification of some streptomycetes producing oxytetracycline

A study was made of Streptomyces rimosus and mutant strains to compare the phenotype of high and low oxytetracycline producers. Strains were identified using a probabilistic identification matrix for the genus Streptomyces. Mutant strains separated into two groups: high-titre strains and blocked mutants. The former identified with the S. rimosus cluster whereas the latter were not identified. Two further oxytetracycline producers identified with the Streptomyces lydicus cluster.     

Numerical classification and identification of Acinetobacter genomic species

A total of 211 Acinetobacter strains (representing all currently recognized genomic species) were tested for 329 biochemical characters. Overall similarities of all strains were determined for 145 characters by numerical taxonomic techniques, the UPGMA algorithm and the S _SM and the S _J coefficients as measures of similarity. Seven clusters (two or more strains) and three unclustered strains were recovered at a similarity level of 80.0% ( S _SM). At this level a complete correspondence between phenotypic cluster and genomic species was found only for genomic species 12 ( Ac. radioresistens ). At higher similarity levels (84.0% to 84.6% ( S _SM)), however, several subclusters were found, each representing a single genomic species. An exception were the strains belonging to the genetically closely related species of the Acinetobacter calcoaceticus-baumannii complex. These were recovered scattered in several subclusters. The degree of genomic relatedness between some DNA groups correlated with phenotypic similarities, especially for DNA group 8 ( Ac. lwoffii ) and 15 of Tjernberg and Ursing, and for DNA group 4 ( Ac. haemolyticus ) and 6.
For the majority of genomic species, two identification matrices were constructed consisting of 22 and 10 diagnostic characters, respectively. The correct identification rates for the matrices were 98.0% (22 tests) and 90.8% (10 tests) taking a Willcox probability >0.9. For unambiguous identification of some genomic species, however, additional methods (preferably DNA-DNA hybridization or ribotyping) should be used.     

Recognition of handwritten similar Chinese characters by self-growing probabilistic decision-based neural network

Recognition of similar (confusion) characters is a difficult problem in optical character recognition (OCR). In this paper, we introduce a neural network solution that is capable of modeling minor differences among similar characters, and is robust to various personal handwriting styles. The Self-growing Probabilistic Decision-based Neural Network (SPDNN) is a probabilistic type neural network, which adopts a hierarchical network structure with nonlinear basis functions and a competitive credit-assignment scheme. Based on the SPDNN model, we have constructed a three-stage recognition system. First, a coarse classifier determines a character to be input to one of the pre-defined subclasses partitioned from a large character set, such as Chinese mixed with alphanumerics. Then a character recognizer determines the input image which best matches the reference character in the subclass. Lastly, the third module is a similar character recognizer, which can further enhance the recognition accuracy among similar or confusing characters. The prototype system has demonstrated a successful application of SPDNN to similar handwritten Chinese recognition for the public database CCL/HCCR1 (5401 characters x200 samples). Regarding performance, experiments on the CCL/HCCR1 database produced 90.12% recognition accuracy with no rejection, and 94.11% accuracy with 6.7% rejection, respectively. This recognition accuracy represents about 4% improvement on the previously announced performance. As to processing speed, processing before recognition (including image preprocessing, segmentation, and feature extraction) requires about one second for an A4 size character image, and recognition consumes approximately 0.27 second per character on a Pentium-100 based personal computer, without use of any hardware accelerator or co-processor.     

Identification and analysis of genetic variation among rose cultivars using random amplified polymorphic DNA

Identified germplasm is an important component for efficient and effective management of plant genetic resources. Traditionally, cultivars or species identification has relied on morphological characters like growth habit or floral morphology like flower colour and other characteristics of the plant. Studies were undertaken for identification and analysis of genetic variation within 34 rose cultivars through random amplified polymorphic DNA (RAPD) markers. Analysis was made by using twenty five decamer primers. Out of twenty five, ten primers were selected and used for identification and analysis of genetic relationships among 34 rose cultivars. A total of 162 distinct DNA fragments ranging from 0.1 to 3.4 kb was amplified by using 10 selected random decamer primers. The genetic similarity was evaluated on the basis of presence or absence of bands. The cluster analysis indicated that the 34 rose cultivars form 9 clusters. The first cluster consists of eight hybrid cultivars, three clusters having five cultivars each, one cluster having four cultivars, two clusters having three cultivars each and two clusters having one cultivar each. The genetic distance was very close within the cultivars. Thus, these RAPD markers have the potential for identification of clusters and characterization of genetic variation within the cultivars. This is also helpful in rose breeding programs and provides a major input into conservation biology.