Jian-Hua Huang

Application of random forests to select premium quality vegetable oils by their fatty acid composition.

In order to discriminate premium quality from inexpensive edible oils, the fatty acid profiles of tea, rapeseed, corn, sunflower and sesame oil were compared with the ones from extra virgin olive oil (EVOO). Fatty acid methyl esters were quantified by GC/MS. Principal component analysis (PCA) and random forests (RF) were applied to cluster the samples. RF showed a better ability of discrimination and also revealed the contribution of each variable to the clustering model. The multidimensional scaling (MDS) plot of the RF proximity matrix demonstrated that tea oil was similar to EVOO. Meanwhile, it was observed that the total content of cis-monounsaturated fatty acids (79.48%) in tea oil was close to EVOO (80.71%), especially the oleic acid (77.38% and 77.45%, respectively). The results suggest that tea oil might be a good edible oil choice, considering the high oleic acid content and similar fatty acid profiles compared to those of EVOO.

Large-scale prediction of drug-target interactions using protein sequences and drug topological structures

The identification of interactions between drugs and target proteins plays a key role in the process of genomic drug discovery. It is both consuming and costly to determine drug-target interactions by experiments alone. Therefore, there is an urgent need to develop new in silico prediction approaches capable of identifying these potential drug-target interactions in a timely manner. In this article, we aim at extending current structure-activity relationship (SAR) methodology to fulfill such requirements. In some sense, a drug-target interaction can be regarded as an event or property triggered by many influence factors from drugs and target proteins. Thus, each interaction pair can be represented theoretically by using these factors which are based on the structural and physicochemical properties simultaneously from drugs and proteins. To realize this, drug molecules are encoded with MACCS substructure fingerings representing existence of certain functional groups or fragments; and proteins are encoded with some biochemical and physicochemical properties. Four classes of drug-target interaction networks in humans involving enzymes, ion channels, G-protein-coupled receptors (GPCRs) and nuclear receptors, are independently used for establishing predictive models with support vector machines (SVMs). The SVM models gave prediction accuracy of 90.31%, 88.91%, 84.68% and 83.74% for four datasets, respectively. In conclusion, the results demonstrate the ability of our proposed method to predict the drug-target interactions, and show a general compatibility between the new scheme and current SAR methodology. They open the way to a host of new investigations on the diversity analysis and prediction of drug-target interactions.

Exploring metabolic syndrome serum profiling based on gas chromatography mass spectrometry and random forest models

Metabolic syndrome (MetS) is a constellation of the most dangerous heart attack risk factors: diabetes and raised fasting plasma glucose, abdominal obesity, high cholesterol and high blood pressure. Analysis and representation of the variances of metabolic profiles is urgently needed for early diagnosis and treatment of MetS. In current study, we proposed a metabolomics approach for analyzing MetS based on GC-MS profiling and random forest models. The serum samples from healthy controls and MetS patients were characterized by GC-MS. Then, random forest (RF) models were used to visually discriminate the serum changes in MetS based on these GC-MS profiles. Simultaneously, some informative metabolites or potential biomarkers were successfully discovered by means of variable importance ranking in random forest models. The metabolites such as 2-hydroxybutyric acid, inositol and d-glucose, were defined as potential biomarkers to diagnose the MetS. These results obtained by proposed method showed that the combining GC-MS profiling with random forest models was a useful approach to analyze metabolites variances and further screen the potential biomarkers for MetS diagnosis.

Classification of Green and Black Teas by PCA and SVM Analysis of Cyclic Voltammetric Signals from Metallic Oxide-Modified Electrode

Forty-three samples of green and black teas were analyzed by an electronic tongue technique. A class of metallic oxide-modified nickel foam electrodes (SnO2, ZnO, TiO2, Bi2O3) was compared in their sensitivity in this system. The signals obtained by cyclic voltammetry were submitted to multivariate data analysis. In the explorative analysis based on principal component analysis (PCA), the score plots showed that two of these sensors were able to distinguish varieties of teas. The resulting PCA scores were modeled with a support vector machine (SVM) that accomplished final prediction with the qualitative classification of teas. The optimal SVM model was achieved after grid search optimization of some parameters and the conduction of the three commonly used kernel functions. With a comparison of classification accuracies, Bi2O3-modified nickel foam electrode performed the best among the four electrodes and SVM model using the polynomial kernel attained the highest within the three used kernels. This work demonstrated that cyclic voltammetry combined with the SVM pattern recognition method could be successfully applied in the classification of green and black teas.

Exploring metabolic syndrome serum free fatty acid profiles based on GC-SIM-MS combined with random forests and canonical correlation analysis.

Metabolic syndrome (MetS) is a cluster of metabolic abnormalities associated with an increased risk of developing cardiovascular diseases or type II diabetes. Till now, the etiology of MetS is complex and still unknown. Metabolic profiling is a powerful tool for exploring metabolic perturbations and potential biomarkers, thus may shed light on the pathophysiological mechanism of diseases. In this study, fatty acid profiling was employed to exploit the metabolic disturbances and discover potential biomarkers of MetS. Fatty acid profiles of serum samples from metabolic syndrome patients and healthy controls were first analyzed by gas chromatography-selected ion monitoring-mass spectrometry (GC-SIM-MS), a robust method for quantitation of fatty acids. Then, the supervised multivariate statistical method of random forests (RF) was used to establish a classification and prediction model for MetS, which could assist the diagnosis of MetS. Furthermore, canonical correlation analysis (CCA) was employed to investigate the relationships between free fatty acids (FFAs) and clinical parameters. As a result, several FFAs, including C16:1n-9c, C20:1n-9c and C22:4n-6c, were identified as potential biomarkers of MetS. The results also indicated that high density lipoprotein-cholesterol (HDL-C), triglycerides (TG) and fasting blood glucose (FBG) were the most important parameters which were closely correlated with FFAs disturbances of MetS, thus they should be paid more attention in clinical practice for monitoring FFAs disturbances of MetS than waist circumference (WC) and systolic blood pressure/diastolic blood pressure (SBP/DBP). The results have demonstrated that metabolic profiling by GC-SIM-MS combined with RF and CCA may be a useful tool for discovering the perturbations of serum FFAs and possible biomarkers for MetS.

Comparison of quantitative structure-retention relationship models on four stationary phases with different polarity for a diverse set of flavor compounds

A quantitative structure-retention relationship study was performed for 656 flavor compounds with highly structural diversity on four stationary phases of different polarities, using topological, constitutional, quantum chemical and geometrical descriptors. Statistical methods were employed to find an informative subset that can accurately predict the gas chromatographic retention indices (RIs). Multivariable linear regression (MLR) was used to map the descriptors to the RIs. The stability and validity of models have been tested by internal and external validation, and good stability and predictive ability were obtained. The resulting QSRR models were well-correlated, with the square of correlation coefficients for cross validation, Q², values of 0.9595, 0.9528, 0.9595 and 0.9223 on stationary phase OV101, DB5, OV17 and C20M, respectively. The molecular properties known to be relevant for GC retention index, such as molecular size, branching, electron density distribution and hydrogen bond effect were well covered by generated descriptors. The descriptors used in models on four stationary phases were compared, and some reasonable explanations about gas chromatographic retention mechanism were obtained. The model may be useful for the prediction of flavor compounds while experimental data is unavailable.

Application of fast Fourier transform cross-correlation and mass spectrometry data for accurate alignment of chromatograms.

Chromatography has been established as one of the most important analytical methods in the modern analytical laboratory. However, preprocessing of the chromatograms, especially peak alignment, is usually a time-consuming task prior to extracting useful information from the datasets because of the small unavoidable differences in the experimental conditions caused by minor changes and drift. Most of the alignment algorithms are performed on reduced datasets using only the detected peaks in the chromatograms, which means a loss of data and introduces the problem of extraction of peak data from the chromatographic profiles. These disadvantages can be overcome by using the full chromatographic information that is generated from hyphenated chromatographic instruments. A new alignment algorithm called CAMS (Chromatogram Alignment via Mass Spectra) is present here to correct the retention time shifts among chromatograms accurately and rapidly. In this report, peaks of each chromatogram were detected based on Continuous Wavelet Transform (CWT) with Haar wavelet and were aligned against the reference chromatogram via the correlation of mass spectra. The aligning procedure was accelerated by Fast Fourier Transform cross correlation (FFT cross correlation). This approach has been compared with several well-known alignment methods on real chromatographic datasets, which demonstrates that CAMS can preserve the shape of peaks and achieve a high quality alignment result. Furthermore, the CAMS method was implemented in the Matlab language and available as an open source package at http://www.github.com/matchcoder/CAMS.

Prediction of retention indices for frequently reported compounds of plant essential oils using multiple linear regression, partial least squares, and support vector machine

Retention indices for frequently reported compounds of plant essential oils on three different stationary phases were investigated. Multivariate linear regression, partial least squares, and support vector machine combined with a new variable selection approach called random-frog recently proposed by our group, were employed to model quantitative structure-retention relationships. Internal and external validations were performed to ensure the stability and predictive ability. All the three methods could obtain an acceptable model, and the optimal results by support vector machine based on a small number of informative descriptors with the square of correlation coefficient for cross validation, values of 0.9726, 0.9759, and 0.9331 on the dimethylsilicone stationary phase, the dimethylsilicone phase with 5% phenyl groups, and the PEG stationary phase, respectively. The performances of two variable selection approaches, random-frog and genetic algorithm, are compared. The importance of the variables was found to be consistent when estimated from correlation coefficients in multivariate linear regression equations and selection probability in model spaces.

Using random forest to classify linear B-cell epitopes based on amino acid properties and molecular features.

Identification and characterization of B-cell epitopes in target antigens was one of the key steps in epitopes-driven vaccine design, immunodiagnostic tests, and antibody production. Experimental determination of epitopes was labor-intensive and expensive. Therefore, there was an urgent need of computational methods for reliable identification of B-cell epitopes. In current study, we proposed a novel peptide feature description method which combined peptide amino acid properties with chemical molecular features. Based on these combined features, a random forest (RF) classifier was adopted to classify B-cell epitopes and non-epitopes. RF is an ensemble method that uses recursive partitioning to generate many trees for aggregating the results; and it always produces highly competitive models. The classification accuracy, sensitivity, specificity, Matthews correlation coefficient (MCC), and area under the curve (AUC) values for current method were 78.31%, 80.05%, 72.23%, 0.5836, and 0.8800, respectively. These results showed that an appropriate combination of peptide amino acid features and chemical molecular features with a RF model could enhance the prediction performance of linear B-cell epitopes. Finally, a freely online service was available at http://sysbio.yznu.cn/Research/Epitopesprediction.aspx.

Using core hydrophobicity to identify phosphorylation sites of human G protein-coupled receptors.

As the most frequent drug target, G protein-coupled receptors (GPCRs) are a large family of seven trans-membrane receptors that sense molecules outside the cell and activate inside signal transduction pathways. The activity and lifetime of activated receptors are regulated by receptor phosphorylation. Therefore, investigating the exact positions of phosphorylation sites in GPCRs sequence could provide useful clues for drug design and other biotechnology applications. Experimental identification of phosphorylation sites is expensive and laborious. Hence, there is significant interest in the development of computational methods for reliable prediction of phosphorylation sites from amino acid sequences. In this article, we presented a simple and effective method to recognize phosphorylation sites of human GPCRs by combining amino acid hydrophobicity and support vector machine. The prediction accuracy, sensitivity, specificity, Matthews correlation coefficient and area under the curve values for phosphoserine, phosphothreonine, and phosphotyrosine were 0.964, 0.790, 0.999, 0.866, 0.941; 0.954, 0.800, 0.985, 0.828, 0.958; and 0.976, 0.820, 0.993, 0.861, 0.959, respectively. The establishment of such a fast and accurate prediction method will speed up the pace of identifying proper GPCRs sites to facilitate drug discovery.