Artificial neural network analysis of pyrolysis mass spectrometric data in the identification of Streptomyces strains

Abstract Sixteen representatives of three morphologically distinct groups of streptomycetes were recovered from soil using selective isolation procedures. Duplicated batches of the test strains were examined by Curie-point pyrolysis mass spectrometry and the first data set used for conventional multivariate statistical analyses and as a training set for an artificial neural network. The second set of data was used for 'operational fingerprinting' and for testing the artificial neural network. All of the test strains were correctly identified using the artificial neural network whereas only fifteen of the sixteen strains were assigned to the correct group using the conventional operational fingerprinting procedure. Artificial neural network analysis of pyrolysis mass spectrometric data provides a rapid, cost-effective and reproducible way of identifying and typing large numbers of microorganisms.     

Resolution of batch variations in pyrolysis mass spectrometry of bacteria by the use of artificial neural network analysis

A simple, but stringent, three group model of bacterial interstrain identity (two cultures of the same strain ofEscherichia coli) and difference (a culture of a serologically distinct strain) was used in multiple serial weekly subcultures for five weeks to demonstrate the effect of both growth-related (phenotypic) and machine-related variation on pyrolysis mass spectra. An aliquot of serum from a single sample was included in each pyrolysis batch to distinguish machine drift from culture drift. Conventional principal component (PC) canonical variate (CV) analysis was successful within each pyrolysis batch but the variations between batches precluded the use of data from more than one batch in successful PCCV analysis. In contrast, artificial neural networks (ANNs) trained with data from one batch could be successfully used to identify groups in data from non-contemporaneous pyrolysis batches. Although the ANN method will require validation in more complex settings than this simple model, it is a promising approach to the problem of batch constraint in pyrolysis mass spectrometry.     

Rapid screening for metabolite overproduction in fermentor broths, using pyrolysis mass spectrometry with multivariate calibration and artificial neural networks

Binary mixtures of model systems consisting of the antibiotic ampicillin with either Escherichia coli or Staphylococcus auresu were subjected to pyrolysis mass spectrometry (PyMS). To deconvolute the pyrolysis mass spectra, so as to obtain quantitative information on the concentration of ampicilin in the mixtures, partial least squares regression (PLS), principal components regression (PCR), and fully interconnected feedforward artificial neural networks (ANNs) were studied. In the latter case, the weights were modified using the standard backpropagation algorithm, and the nodes used a sigmoidal squsahing funciton. It was found that each of the methods could be used to provide calibration models which gave excellent predictions for the concentrations of ampicillin in samples on which they had not been trained. Furthermore, ANNs trained to predict the amount of ampicilin in E. coli were able to generalise so as to predict the concentration of ampicillin in a S. aureus background, illustrating the robustness of ANNs to rather substantial variations in the biological background. The PyMS of the complex mixture of ampicilin in bacteria could not be expressed simply in terms of additive combinations of the spectra describing the pure components of the mixtures and their relative concentrations. Intermolecular reactions took place in the pyrolysate, leading to a lack of superposition of the spectral components and to a dependence of the normalized mass spectrum on sample size. Samples from fermentations of a single organism in a complex production medium were also analyzed quantitatively for a drug of commercial interest. The drug could also be quantified in a variety of mutant-producing strains cultivated in the same medium. The combination of PyMS and ANNs constitutes a novel, rapid, and convenient method for exploitation in strain improvement screening programs. (c) 1994 John Wiley & Sons, Inc.     

Prediction of siRNA knockdown efficiency using artificial neural network models

Selective knockdown of gene expression by short interference RNAs (siRNAs) has allowed rapid validation of gene functions and made possible a high throughput, genome scale approach to interrogate gene function. However, randomly designed siRNAs display different knockdown efficiencies of target genes. Hence, various prediction algorithms based on siRNA functionality have recently been constructed to increase the likelihood of selecting effective siRNAs, thereby reducing the experimental cost. Toward this end, we have trained three Back-propagation and Bayesian neural network models, previously not used in this context, to predict the knockdown efficiencies of 180 experimentally verified siRNAs on their corresponding target genes. Using our input coding based primarily on RNA structure thermodynamic parameters and cross-validation method, we showed that our neural network models outperformed most other methods and are comparable to the best predicting algorithm thus far published. Furthermore, our neural network models correctly classified 74% of all siRNAs into different efficiency categories; with a correlation coefficient of 0.43 and receiver operating characteristic curve score of 0.78, thus highlighting the potential utility of this method to complement other existing siRNA classification and prediction schemes.     

Genus-wide Bacillus species identification through proper artificial neural network experiments on fatty acid profiles

Gas chromatographic fatty acid methyl ester analysis of bacteria is an easy, cheap and fast-automated identification tool routinely used in microbiological research. This paper reports on the application of artificial neural networks for genus-wide FAME-based identification of Bacillus species. Using 1,071 FAME profiles covering a genus-wide spectrum of 477 strains and 82 species, different balanced and imbalanced data sets have been created according to different validation methods and model parameters. Following training and validation, each classifier was evaluated on its ability to identify the profiles of a test set. Comparison of the classifiers showed a good identification rate favoring the imbalanced data sets. The presence of the Bacillus cereus and Bacillus subtilis groups made clear that it is of great importance to take into account the limitations of FAME analysis resolution for the construction of identification models. Indeed, as members of such a group cannot easily be distinguished from one another based upon FAME data alone, identification models built upon this data can neither be successful at keeping them apart. Comparison of the different experimental setups ultimately led to a few general recommendations. With respect to the routinely used commercial Sherlock Microbial Identification System (MIS, Microbial ID, Inc. (MIDI), Newark, Delaware, USA), the artificial neural network test results showed a significant improvement in Bacillus species identification. These results indicate that machine learning techniques such as artificial neural networks are most promising tools for FAME-based classification and identification of bacterial species.     

Cell-cycle-dependent variations in FTIR micro-spectra of single proliferating HeLa cells: Principal component and artificial neural network analysis

We have previously reported spectral differences for cells at different stages of the eukaryotic cell division cycle. These differences are due to the drastic biochemical and morphological changes that occur as a consequence of cell proliferation. We correlate these changes in FTIR absorption and Raman spectra of individual cells with their biochemical age (or phase in the cell cycle), determined by immunohistochemical staining to detect the appearance (and subsequent disappearance) of cell-cycle-specific cyclins, and/or the occurrence of DNA synthesis. Once spectra were correlated with their cells' staining patterns, we used methods of multivariate statistics to analyze the changes in cellular spectra as a function of cell cycle phase.     

Rapid identification using pyrolysis mass spectrometry and artificial neural networks of Propionibacterium acnes isolated from dogs

Curie-point pyrolysis mass spectra were obtained from reference Propionibacterium strains and canine isolates. Artificial neural networks (ANNs) were trained by supervised learning (with the back-propagation algorithm) to recognize these strains from their pyrolysis mass spectra; all the strains isolated from dogs were identified as human wild type P. acnes. This is an important nosological discovery, and demonstrates that the combination of pyrolysis mass spectrometry and ANNs provides an objective, rapid and accurate identification technique. Bacteria isolated from different biopsy specimens from the same dog were found to be separate strains of P. acnes , demonstrating a within-animal variation in microflora. The classification of the canine isolates by Kohonen artificial neural networks (KANNs) was compared with the classical multivariate techniques of canonical variates analysis and hierarchical cluster analysis, and found to give similar results. This is the first demonstration, within microbiology, of KANNs as an unsupervised clustering technique which has the potential to group pyrolysis mass spectra both automatically and relatively objectively.     

Recognition of culture state using two-dimensional gel electrophoresis with an artificial neural network

Proteomic technologies were applied to the examination of nutrient components in culture broth. In bioprocesses, many types of media have been proposed and used on the commercial scale. Natural nutrients, the chemical components of which cannot be identified completely, are often used in fermentation processes such as in the production of baker's yeast, alcoholic beverages, amino acids, and pharmaceuticals. The catabolic activities of the microorganisms in these processes vary with the species used. We used an artificial neural network (ANN) to recognize the sufficiency of chemical elements based on the protein spots resolved in 2-DE, and we evaluated this technique using the leave-one-out method. We also attempted to reduce the number of input data for spot selection based on sensitivity analysis of the ANN, and the selected data were used to improve accuracy.     

Study of the restitution of action potential duration using the artificial neural network

It is widely accepted that the APD (action potential duration) restitution plays a key role in the initializing and maintaining of the reentry arrhythmias. The Luo-Rudy II models paced with different protocols showed that the current APD had a complex relation with the previous APDs and diastole intervals (DIs). This relation could not be accurately described by a single exponential function. We used an artificial neural network to formularize this relation. The results suggested that back-propagation (BP) network could predict the current APD from the information of the first three previous beats. This would help provide a target for potential anti-arrhythmic therapies.     

Rapid identification of an antibody DNA construct rearrangement sequence variant by mass spectrometry

During cell line development for an IgG1 antibody candidate (mAb1), a C-terminal extension was identified in 2 product candidate clones expressed in CHO-K1 cell line. The extension was initially observed as the presence of anomalous new peaks in these clones after analysis by cation exchange chromatography (CEX-HPLC) and reduced capillary electrophoresis (rCE-SDS). Reduced mass analysis of these CHO-K1 clones revealed that a larger than expected mass was present on a sub-population of the heavy chain species, which could not be explained by any known chemical or post-translational modifications. It was suspected that this additional mass on the heavy chain was due to the presence of an additional amino acid sequence. To identify the suspected additional sequence, de novo sequencing in combination with proteomic searching was performed against translated DNA vectors for the heavy chain and light chain. Peptides unique to the clones containing the extension were identified matching short sequences (corresponding to 9 and 35 amino acids, respectively) from 2 non-coding sections of the light chain vector construct. After investigation, this extension was observed to be due to the re-arrangement of the DNA construct, with the addition of amino acids derived from the light chain vector non-translated sequence to the C-terminus of the heavy chain. This observation showed the power of proteomic mass spectrometric techniques to identify an unexpected antibody sequence variant using de novo sequencing combined with database searching, and allowed for rapid identification of the root cause for new peaks in the cation exchange and rCE-SDS assays.     

Rapid identification of an antibody DNA construct rearrangement sequence variant by mass spectrometry

During cell line development for an IgG1 antibody candidate (mAb1), a C-terminal extension was identified in 2 product candidate clones expressed in CHO-K1 cell line. The extension was initially observed as the presence of anomalous new peaks in these clones after analysis by cation exchange chromatography (CEX-HPLC) and reduced capillary electrophoresis (rCE-SDS). Reduced mass analysis of these CHO-K1 clones revealed that a larger than expected mass was present on a sub-population of the heavy chain species, which could not be explained by any known chemical or post-translational modifications. It was suspected that this additional mass on the heavy chain was due to the presence of an additional amino acid sequence. To identify the suspected additional sequence, de novo sequencing in combination with proteomic searching was performed against translated DNA vectors for the heavy chain and light chain. Peptides unique to the clones containing the extension were identified matching short sequences (corresponding to 9 and 35 amino acids, respectively) from 2 non-coding sections of the light chain vector construct. After investigation, this extension was observed to be due to the re-arrangement of the DNA construct, with the addition of amino acids derived from the light chain vector non-translated sequence to the C-terminus of the heavy chain. This observation showed the power of proteomic mass spectrometric techniques to identify an unexpected antibody sequence variant using de novo sequencing combined with database searching, and allowed for rapid identification of the root cause for new peaks in the cation exchange and rCE-SDS assays.     

应用人工神经网络评价湖泊的富营养化

应用人工神经网络方法,以化学需氧量、总氮、总磷和透明度作为评价参数,经反复尝试,构建了具有４层结构用于评价湖泊富营养化的误差逆传播网络．其输入层有４个神经元,２个隐含层也各有４个神经元,输出层有１个神经元．以太湖富营养化评价标准作为样本模式提供给网络,按照误差逆传播网络的学习规则对网络进行训练,经过３７６８４次学习后,网络达到预先给定的收敛标准．使网络具备了识别湖泊富营养化程度的功能．应用该网络对我国１７个湖泊的富营养化程度进行评价,操作过程简便易行,评价结果切合实际,展示了这种方法的一系列优点．     

Differences in relative growth rate in 11 grasses correlate with differences in chemical composition as determined by pyrolysis mass spectrometry

Summary Eleven grass species varying in potential relative growth rate (RGR) were investigated for differences in chemical composition by pyrolysis mass spectrometry. The spectral data revealed correlations between RGR and the relative composition of several biopolymers. Species with a low potential RGR contained relatively more cell wall material such as lignin, hemicellulose, cellulose, polysaccharide-bound ferulic acid and hydroxyproline-rich protein, whereas species with a high potential RGR showed relatively more cytoplasmic elements such as protein (other than those incorporated in cell walls) and sterols.     

BP神经网络优化微生物浸矿工艺

以低品位黄铜矿溶液为原料,浸矿制备Cu2＋能有效提高低品位黄铜矿的利用价值。基于浸矿过程中存在多因素影响的现象,通过正交试验与神经网络分析方法,对浸矿条件（接种量、矿石品位、Fe2＋添加量及浸矿溶液pH）实行优化。结果表明：在正交试验组中最佳试验结果为浸矿产128．753mg／LCu^2＋;BP神经网络优化后的最佳实验组合为微生物接种量12％、矿石品位0．3％、添加Fe^2＋24g/L及浸矿溶液pH1．7,该条件下验证试验产Cu^2＋ 141．352mg／L,通过正交试验及神经网络优化提高了微生物浸出低品位黄铜矿酸性溶液Cu^2＋的产量。     

基质辅助激光解吸电离飞行时间质谱技术用于常见益生菌鉴定的初步研究

目的建立基质辅助激光解吸电离飞行时间质谱(MADLI-TOF MS)技术鉴定常见益生菌的实验方法并对MADLI-TOF MS技术的适用性进行初步评价。方法对MADLI-TOF MS技术鉴定常见益生菌过程中各影响因素进行考察,筛选出最佳的实验条件。利用19株供试菌株所得的蛋白指纹图谱对MADLI-TOF MS技术的适用性进行研究。结果建立了MADLI-TOF MS技术鉴定常见益生菌的最佳实验方法。初步证明MADLI-TOF MS技术具备在属、种、亚种以及菌株水平上鉴定常见益生菌的能力。结论建立的实验方法稳定性高、重复性好,可以作为MADLI-TOF MS技术鉴定常见益生菌的参考方法。MADLI-TOF MS技术可以作为常见益生菌鉴定的方法之一。     

Rapid phenotypic characterization of Salmonella enterica strains by pyrolysis metastable atom bombardment mass spectrometry with multivariate statistical and artificial neural network pattern recognition

Pyrolysis mass spectrometry was investigated for rapid characterization of bacteria. Spectra of Salmonella were compared to their serovars, pulsed-field gel electrophoresis (PFGE) patterns, antibiotic resistance profiles, and MIC values. Pyrolysis mass spectra generated via metastable atom bombardment were analyzed by multivariate principal component-discriminant analysis and artificial neural networks (ANNs). Spectral patterns developed by discriminant analysis and tested with Leave-One-Out (LOO) cross-validation distinguished Salmonella strains by serovar (97% correct) and by PFGE groups (49%). An ANN model of the same PFGE groups was cross-validated, using the LOO rule, with 92% agreement. Using an ANN, thirty previously unseen spectra were correctly classified by serotype (97%) and at the PFGE level (67%). Attempts by ANN to model spectra grouped by resistance profile-but ignoring PFGE or serotype-failed (10% correct), but ANNs differentiating ten samples of the same serotype/PFGE class were more successful. To assess the information content of PyMS data serendipitously associated with or directly related to resistance character, the ten isolates were grouped into four, three, or two categories. The four categories corresponded to four resistance profiles. The four class and three class ANNs showed much improved but insufficient modeling power. The two-class ANN and a corresponding multivariate model maximized inferential power for a coarse antibiotic-resistance-related distinction. They each cross-validated by LOO at 90%. This is the first direct correlation of pyrolysis metastable atom bombardment mass spectrometry with immunological (e.g. serology) or molecular biology (e.g. PFGE) based techniques.     

QSAR study of the enantiomeric excess in asymmetric catalytic reactions with topological indices and an artificial neural network

The relationships between the enantiomer excess of product in catalytic asymmetric reactions and the structures of the catalysts or reagents in several asymmetric reactions were studied using a backpropagation (BP) neural network with topological indices and their chiral expansions. The trained network can be used to screen new asymmetric catalysts, estimate catalytic effects, design reaction environments, and prove or improve the proposed reaction mechanism.     

Rapid analysis of O-acetylated neuraminic acids by matrix assisted laser desorption/ionization time-of-flight mass spectrometry

N-Acetylneuraminic acid (a sialic acid) occurs mainly as a terminal substituent of oligosaccharides of glycoconjugates. Derivatives of neuraminic acid occur widely, substituted in the amino and hydroxy side chains, as well in the C-9 carbon skeleton. These derivatives are responsible for specific functions of sialic acids during cell-cell, cell-substrate, or cell-virus interactions. The study of O-acetylated neuraminic acids is difficult, because only small amounts are extractable from natural sources and they are generally unstable to acids and bases. We report a new method for the rapid analysis of O-acetylated neuraminic acids, using a combination of reversed phase HPLC and MALDI-TOF mass spectrometry. A mixture of neuraminic acids from bovine submaxillary gland mucins was analysed, as well as neuraminic acids variously substituted in the amino and hydroxy side chains with acetyl and glycolyl groups, respectively. © 1998 Rapid Science Ltd     

Differentiation of Listeria monocytogenes serovars by using artificial neural network analysis of Fourier-transformed infrared spectra

A classification system based on Fourier transform infrared (FTIR) spectroscopy combined with artificial neural network analysis was designed to differentiate 12 serovars of Listeria monocytogenes using a reference database of 106 well-defined strains. External validation was performed using a test set of another 166 L. monocytogenes strains. The O antigens (serogroup) of 164 strains (98.8%) could be identified correctly, and H antigens were correctly determined in 152 (91.6%) of the test strains. Importantly, 40 out of 41 potentially epidemic serovar 4b strains were unambiguously identified. FTIR analysis is superior to PCR-based systems for serovar differentiation and has potential for the rapid, simultaneous identification of both species and serovar of an unknown Listeria isolate by simply measuring a whole-cell infrared spectrum.