Abel Folch-Fortuny

Calibration transfer between NIR spectrometers: New proposals and a comparative study

Calibration transfer between near‐infrared (NIR) spectrometers is a subtle issue in chemometrics and process industry. In fact, as even very similar instruments may generate strongly different spectral responses, regression models developed on a first NIR system can rarely be used with spectra collected by a second apparatus. In this work, two novel methods to perform calibration transfer between NIR spectrometers are proposed. Both of them permit to exploit the specific relationships between instruments for imputing new unmeasured spectra, which will be then resorted to for building an improved predictive model, suitable for the analysis of future incoming data. Specifically, the two approaches are based on trimmed scores regression and joint‐Y partial least squares regression, respectively. The performance of these novel strategies will be assessed and compared to that of well‐established techniques such as maximum likelihood principal component analysis and piecewise direct standardisation in two real case studies.

Topology analysis and visualization of Potyvirus protein-protein interaction network

BackgroundOne of the central interests of Virology is the identification of host factors that contribute to virus infection. Despite tremendous efforts, the list of factors identified remains limited. With omics techniques, the focus has changed from identifying and thoroughly characterizing individual host factors to the simultaneous analysis of thousands of interactions, framing them on the context of protein-protein interaction networks and of transcriptional regulatory networks. This new perspective is allowing the identification of direct and indirect viral targets. Such information is available for several members of the Potyviridae family, one of the largest and more important families of plant viruses.ResultsAfter collecting information on virus protein-protein interactions from different potyviruses, we have processed it and used it for inferring a protein-protein interaction network. All proteins are connected into a single network component. Some proteins show a high degree and are highly connected while others are much less connected, with the network showing a significant degree of dissortativeness. We have attempted to integrate this virus protein-protein interaction network into the largest protein-protein interaction network of Arabidopsis thaliana, a susceptible laboratory host. To make the interpretation of data and results easier, we have developed a new approach for visualizing and analyzing the dynamic spread on the host network of the local perturbations induced by viral proteins. We found that local perturbations can reach the entire host protein-protein interaction network, although the efficiency of this spread depends on the particular viral proteins. By comparing the spread dynamics among viral proteins, we found that some proteins spread their effects fast and efficiently by attacking hubs in the host network while other proteins exert more local effects.ConclusionsOur findings confirm that potyvirus protein-protein interaction networks are highly connected, with some proteins playing the role of hubs. Several topological parameters depend linearly on the protein degree. Some viral proteins focus their effect in only host hubs while others diversify its effect among several proteins at the first step. Future new data will help to refine our model and to improve our predictions.

Enabling network inference methods to handle missing data and outliers

BackgroundThe inference of complex networks from data is a challenging problem in biological sciences, as well as in a wide range of disciplines such as chemistry, technology, economics, or sociology. The quantity and quality of the data greatly affect the results. While many methodologies have been developed for this task, they seldom take into account issues such as missing data or outlier detection and correction, which need to be properly addressed before network inference.ResultsHere we present an approach to (i) handle missing data and (ii) detect and correct outliers based on multivariate projection to latent structures. The method, called trimmed scores regression (TSR), enables network inference methods to analyse incomplete datasets by imputing the missing values coherently with the latent data structure. Furthermore, it substitutes the faulty values in a dataset by proper estimations. We provide an implementation of this approach, and show how it can be integrated with any network inference method as a preliminary data curation step. This functionality is demonstrated with a state of the art network inference method based on mutual information distance and entropy reduction, MIDER.ConclusionThe methodology presented here enables network inference methods to analyse a large number of incomplete and faulty datasets that could not be reliably analysed so far. Our comparative studies show the superiority of TSR over other missing data approaches used by practitioners. Furthermore, the method allows for outlier detection and correction.

Assessment of maximum likelihood PCA missing data imputation

Maximum likelihood principal component analysis (MLPCA) was originally proposed to incorporate measurement error variance information in principal component analysis (PCA) models. MLPCA can be used to fit PCA models in the presence of missing data, simply by assigning very large variances to the non‐measured values. An assessment of maximum likelihood missing data imputation is performed in this paper, analysing the algorithm of MLPCA and adapting several methods for PCA model building with missing data to its maximum likelihood version. In this way, known data regression (KDR), KDR with principal component regression (PCR), KDR with partial least squares regression (PLS) and trimmed scores regression (TSR) methods are implemented within the MLPCA method to work as different imputation steps. Six data sets are analysed using several percentages of missing data, comparing the performance of the original algorithm, and its adapted regression‐based methods, with other state‐of‐the‐art methods. Copyright

PLS model building with missing data: New algorithms and a comparative study

New algorithms to deal with missing values in predictive modelling are presented in this article. Specifically, 2 trimmed scores regression adaptations are proposed, one from principal component analysis model building with missing data (MD) and other from partial least squares regression model exploitation with missing values. Using these methods, practitioners can impute MD both in the explanatory/predictor and the dependent/response variables. Partial least squares is used here to build the multivariate calibration models; however, any regression method can be used after MD imputation. Four case studies, with different latent structures, are analysed here to compare the trimmed scores regression–based methods against state‐of‐the‐art approaches. The MATLAB code for these methods is also provided for its direct implementation at http://mseg.webs.upv.es, under a GNU license.

Dynamic elementary mode modelling of non-steady state flux data

BackgroundA novel framework is proposed to analyse metabolic fluxes in non-steady state conditions, based on the new concept of dynamic elementary mode (dynEM): an elementary mode activated partially depending on the time point of the experiment.ResultsTwo methods are introduced here: dynamic elementary mode analysis (dynEMA) and dynamic elementary mode regression discriminant analysis (dynEMR-DA). The former is an extension of the recently proposed principal elementary mode analysis (PEMA) method from steady state to non-steady state scenarios. The latter is a discriminant model that permits to identify which dynEMs behave strongly different depending on the experimental conditions. Two case studies of Saccharomyces cerevisiae, with fluxes derived from simulated and real concentration data sets, are presented to highlight the benefits of this dynamic modelling.ConclusionsThis methodology permits to analyse metabolic fluxes at early stages with the aim of i) creating reduced dynamic models of flux data, ii) combining many experiments in a single biologically meaningful model, and iii) identifying the metabolic pathways that drive the organism from one state to another when changing the environmental conditions.

Fusion of genomic, proteomic and phenotypic data: the case of potyviruses

Data fusion has been widely applied to analyse different sources of information, combining all of them in a single multivariate model. This methodology is mandatory when different omic data sets must be integrated to fully understand an organism using a systems biology approach. Here, a data fusion procedure is presented to combine genomic, proteomic and phenotypic data sets gathered for Tobacco etch virus (TEV). The genomic data correspond to random mutations inserted in most viral genes. The proteomic data represent both the effect of these mutations on the encoded proteins and the perturbation induced by the mutated proteins to their neighbours in the protein-protein interaction network (PPIN). Finally, the phenotypic trait evaluated for each mutant virus is replicative fitness. To analyse these three sources of information a Partial Least Squares (PLS) regression model is fitted in order to extract the latent variables from data that explain (and relate) the significant variables to the fitness of TEV. The final output of this methodology is a set of functional modules of the PPIN relating topology and mutations with fitness. Throughout the re-analysis of these diverse TEV data, we generated valuable information on the mechanism of action of certain mutations and how they translate into organismal fitness. Results show that the effect of some mutations goes beyond the protein they directly affect and spreads on the PPIN to neighbour proteins, thus defining functional modules.