Andreas Rytz

The limit fold change model: A practical approach for selecting differentially expressed genes from microarray data

BackgroundThe biomedical community is developing new methods of data analysis to more efficiently process the massive data sets produced by microarray experiments. Systematic and global mathematical approaches that can be readily applied to a large number of experimental designs become fundamental to correctly handle the otherwise overwhelming data sets.ResultsThe gene selection model presented herein is based on the observation that: (1) variance of gene expression is a function of absolute expression; (2) one can model this relationship in order to set an appropriate lower fold change limit of significance; and (3) this relationship defines a function that can be used to select differentially expressed genes. The model first evaluates fold change (FC) across the entire range of absolute expression levels for any number of experimental conditions. Genes are systematically binned, and those genes within the top X% of highest FCs for each bin are evaluated both with and without the use of replicates. A function is fitted through the top X% of each bin, thereby defining a limit fold change. All genes selected by the 5% FC model lie above measurement variability using a within standard deviation (SDwithin) confidence level of 99.9%. Real time-PCR (RT-PCR) analysis demonstrated 85.7% concordance with microarray data selected by the limit function.ConclusionThe FC model can confidently select differentially expressed genes as corroborated by variance data and RT-PCR. The simplicity of the overall process permits selecting model limits that best describe experimental data by extracting information on gene expression patterns across the range of expression levels. Genes selected by this process can be consistently compared between experiments and enables the user to globally extract information with a high degree of confidence.

When machine tastes coffee: instrumental approach to predict the sensory profile of espresso coffee.

A robust and reproducible model was developed to predict the sensory profile of espresso coffee from instrumental headspace data. The model is derived from 11 different espresso coffees and validated using 8 additional espressos. The input of the model consists of (i) sensory profiles from a trained panel and (ii) on-line proton-transfer reaction mass spectrometry (PTR-MS) data. The experimental PTR-MS conditions were designed to simulate those for the sensory evaluation. Sixteen characteristic ion traces in the headspace were quantified by PTR-MS, requiring only 2 min of headspace measurement per espresso. The correlation is based on a knowledge-based standardization and normalization of both datasets that selectively extracts differences in the quality of samples, while reducing the impact of variations on the overall intensity of coffees. This work represents a significant progress in terms of correlation of sensory with instrumental results exemplified on coffee.

Training is a critical step to obtain reliable product profiles in a real food industry context

In the recent past, many authors have evaluated the benefit of training in the case of descriptive profiling. Conflicting results were presented, sometimes even concluding that untrained panels performed as well as trained ones. In this study, a panel of ten assessors evaluated eight soluble coffees from a benchmarking with repetition before and after a training of 21 h using a pre-defined glossary of 17 attributes. The benefits of training are multiple in this typical example from the food industry, where complex products are described within a relatively narrow sensory range. Two other means are proposed to shorten the duration of the study without altering the relevance of the data. First, monitoring panel performance during training helps to keep the training duration as short as possible. Second, a single evaluation of products is enough to provide reliable data for products as homogeneous as soluble coffee. In other more heterogeneous products, such as meat, cheese, etc., replicates are essential in order to determine the variability within products with the same treatment or even with the same product.

Microarray data analysis: a practical approach for selecting differentially expressed genes

BackgroundThe biomedical community is rapidly developing new methods of data analysis for microarray experiments, with the goal of establishing new standards to objectively process the massive datasets produced from functional genomic experiments. Each microarray experiment measures thousands of genes simultaneously producing an unprecedented amount of biological information across increasingly numerous experiments; however, in general, only a very small percentage of the genes present on any given array are identified as differentially regulated. The challenge then is to process this information objectively and efficiently in order to obtain knowledge of the biological system under study and by which to compare information gained across multiple experiments. In this context, systematic and objective mathematical approaches, which are simple to apply across a large number of experimental designs, become fundamental to correctly handle the mass of data and to understand the true complexity of the biological systems under study.ResultsThe present report develops a method of extracting differentially expressed genes across any number of experimental samples by first evaluating the maximum fold change (FC) across all experimental parameters and across the entire range of absolute expression levels. The model developed works by first evaluating the FC across the entire range of absolute expression levels in any number of experimental conditions. The selection of those genes within the top X% of highest FCs observed within absolute expression bins was evaluated both with and without the use of replicates. Lastly, the FC model was validated by both real time polymerase chain reaction (RT-PCR) and variance data. Semi-quantitative RT-PCR analysis demonstrated 73% concordance with the microarray data from Mu11K Affymetrix GeneChips. Furthermore, 94.1% of those genes selected by the 5% FC model were found to lie above measurement variability using a SDwithin confidence level of 99.9%.ConclusionAs evidenced by the high rate of validation, the FC model has the potential to minimize the number of required replicates in expensive microarray experiments by extracting information on gene expression patterns (e.g. characterizing biological and/or measurement variance) within an experiment. The simplicity of the overall process allows the analyst to easily select model limits which best describe the data. The genes selected by this process can be compared between experiments and are shown to objectively extract information which is biologically & statistically significant.

The Global Error Assessment (GEA) model for the selection of differentially expressed genes in microarray data

MOTIVATION Microarray technology has become a powerful research tool in many fields of study; however, the cost of microarrays often results in the use of a low number of replicates (k). Under circumstances where k is low, it becomes difficult to perform standard statistical tests to extract the most biologically significant experimental results. Other more advanced statistical tests have been developed; however, their use and interpretation often remain difficult to implement in routine biological research. The present work outlines a method that achieves sufficient statistical power for selecting differentially expressed genes under conditions of low k, while remaining as an intuitive and computationally efficient procedure. RESULTS The present study describes a Global Error Assessment (GEA) methodology to select differentially expressed genes in microarray datasets, and was developed using an in vitro experiment that compared control and interferon-gamma treated skin cells. In this experiment, up to nine replicates were used to confidently estimate error, thereby enabling methods of different statistical power to be compared. Gene expression results of a similar absolute expression are binned, so as to enable a highly accurate local estimate of the mean squared error within conditions. The model then relates variability of gene expression in each bin to absolute expression levels and uses this in a test derived from the classical ANOVA. The GEA selection method is compared with both the classical and permutational ANOVA tests, and demonstrates an increased stability, robustness and confidence in gene selection. A subset of the selected genes were validated by real-time reverse transcription-polymerase chain reaction (RT-PCR). All these results suggest that GEA methodology is (i) suitable for selection of differentially expressed genes in microarray data, (ii) intuitive and computationally efficient and (iii) especially advantageous under conditions of low k. AVAILABILITY The GEA code for R software is freely available upon request to authors.

The proportion of lycopene isomers in human plasma is modulated by lycopene isomer profile in the meal but not by lycopene preparation.

Dietary lycopene consists mostly of the (all-E) isomer. Upon absorption, (all-E) lycopene undergoes isomerisation into various (Z)-isomers. Because these isomers offer potentially better health benefits than the (all-E) isomer, the aim of the present study was to investigate if the profile of lycopene isomers in intestinal lipoproteins is affected by the profile of lycopene isomers in the meal and by the tomato preparation. Six postprandial, crossover tests were performed in healthy men. Three meals provided about 70 % of the lycopene as (Z)-isomers, either mainly as 5-(Z) or 13-(Z), or as a mixture of 9-(Z) and 13-(Z) lycopene, while three tomato preparations provided lycopene mainly as the (all-E) isomer. Consumption of the 5-(Z) lycopene-rich meal led to a high (60 %) proportion of this isomer in TAG-rich lipoproteins (TRL), indicating a good absorption and/or a low intestinal conversion of this isomer. By contrast, consumption of meals rich in 9-(Z) and 13-(Z) lycopene isomers resulted in a low level of these isomers but high amounts of the 5-(Z) and (all-E) isomers in TRL. This indicates that the 9-(Z) and 13-(Z) isomers were less absorbed or were converted into 5-(Z) and (all-E) isomers. Dietary (Z)-lycopene isomers were, therefore, differently isomerised and released in TRL during their intestinal absorption in men. Consuming the three meals rich in (all-E) lycopene resulted in similar proportions of lycopene isomers in TRL: 60 % (all-E), 20 % 5-(Z), 9 % 13-(Z), 2 % 9-(Z) and 9 % unidentified (Z)-isomers. These results show that the tomato preparation has no impact on the lycopene isomerisation occurring during absorption in humans.

Non-covalent binding of proteins to polyphenols correlates with their amino acid sequence

The present paper describes the assessment of non-covalent binding (NCB) between milk proteins and polyphenols and its correlation with the physicochemical parameters of proteins. A method based on ultrafiltration and liquid chromatography-tandem mass spectrometry was used to analyse free and non-covalently bound polyphenols (ligands) in mixtures with major milk proteins. Binding strength values of individual polyphenols were normalised to those obtained with quercitrin (quercetin-3-O-rhamnoside), used as a reference compound. NCB data acquired by experiments at pH 6.6 without any preliminary protein denaturation were correlated with the physicochemical parameters of ligands and proteins. Unsupervised multivariate analysis revealed that NCB of proteins clustered according to their family (caseins separated from albumins). Based on this model, a predictive relationship was observed between protein-polyphenol binding strength and primary/secondary structure parameters of the proteins e.g. number of charges, proline residues and extended strand. These results confirm that, under the investigated experimental conditions, the NCB between polyphenols and protein mixtures can be predicted and optimised based on the molecular structures.

Dietary fat induces sustained reward response in the human brain without primary taste cortex discrimination.

To disentangle taste from reward responses in the human gustatory cortex, we combined high density electro-encephalography with a gustometer delivering tastant puffs to the tip of the tongue. Stimuli were pure tastants (salt solutions at two concentrations), caloric emulsions (two milk preparations identical in composition except for fat content) and a mixture of high fat milk with the lowest salt concentration. Early event-related potentials (ERPs) showed a dose-response effect for increased taste intensity, with higher amplitude and shorter latency for high compared to low salt concentration, but not for increased fat content. However, the amplitude and distribution of late potentials were modulated by fat content independently of reported intensity and discrimination. Neural source estimation revealed a sustained activation of reward areas to the two high-fat stimuli. The results suggest calorie detection through specific sensors on the tongue independent of perceived taste. Finally, amplitude variation of the first peak in the event-related potential to the different stimuli correlated with papilla density, suggesting a higher discrimination power for subjects with more fungiform papillae.

Generation of 4-Hydroxy-2,5-Dimethyl-3(2H)-Furanone from Rhamnose as Affected by Reaction Parameters: Experimental Design Approach

The formation of 4-hydroxy-2,5-dimethyl-3(2H)-furanone (HDMF) was studied in aqueous model systems containing L-rhamnose and L-lysine. The approach consisted in systematically varying four reaction parameters (rhamnose concentration, rhamnose to lysine ratio, pH, and phosphate concentration) at 3 levels. A fractional factorial design was used to reduce the number of trials. The degradation of rhamnose was followed by high performance anion exchange chromatography and the formation of HDMF by solid phase extraction in combination with GC/MS. The study permitted the identification of critical reaction parameters that affect the formation of HDMF from rhamnose in aqueous systems. Although all studied parameters have some impact on the HDMF formation and rhamnose degradation kinetics, the effect of phosphate is by far the most important, followed by concentration of precursors and pH. The experimental design approach permitted us, with a limited number of experiments, to accurately model the effects of the four investigated reaction parameters on the kinetics of rhamnose degradation and HDMF formation (R(2)>0.93). Overall, the results indicate that rhamnose can be an excellent precursor of HDMF (yield >40 mol%), if the reaction conditions are well mastered.

Is portion size selection associated with expected satiation, perceived healthfulness or expected tastiness? A case study on pizza using a photograph-based computer task

Increasing portion sizes over the last 30 years are considered to be one of the factors underlying overconsumption. Past research on the drivers of portion selection for foods showed that larger portions are selected for foods delivering low expected satiation. However, the respective contribution of expected satiation vs. two other potential drivers of portion size selection, i.e. perceived healthfulness and expected tastiness, has never been explored. In this study, we conjointly explored the role of expected satiation, perceived healthfulness and expected tastiness when selecting portions within a range of six commercial pizzas varying in their toppings and brands. For each product, 63 pizza consumers selected a portion size that would satisfy them for lunch and scored their expected satiation, perceived healthfulness and expected tastiness. As six participants selected an entire pizza as ideal portion independently of topping or brand, their data sets were not considered in the data analyses completed on responses from 57 participants. Hierarchical multiple regression analyses showed that portion size variance was predicted by perceived healthiness and expected tastiness variables. Two sub-groups of participants with different portion size patterns across pizzas were identified through post-hoc exploratory analysis. The explanatory power of the regression model was significantly improved by adding interaction terms between sub-group and expected satiation variables and between sub-group and perceived healthfulness variables to the model. Analysis at a sub-group level showed either positive or negative association between portion size and expected satiation depending on sub-groups. For one group, portion size selection was more health-driven and for the other, more hedonic-driven. These results showed that even when considering a well-liked product category, perceived healthfulness can be an important factor influencing portion size decision.