Search results

The 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. Results Here 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. Conclusion The 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. ; Research in this study was partially supported by the European Union through project BioPreDyn (FP7-KBBE 289434), and the Spanish Ministry of Science and Innovation and FEDER funds from the European Union through grants MultiScales (DPI2011-28112-C04-02, DPI2011-28112-C04-03), and SynBioFactory (DPI2014-55276-C5-1-R, DPI2014-55276-C5-2-R). AF Villaverde also acknowledges funding from the Xunta de Galicia through an I2C postdoctoral fellowship (I2C ED481B 2014/133-0). We also ...

Signaling pathways play a key role in complex diseases such as cancer, for which the development of novel therapies is a difficult, expensive and laborious task. Computational models that can predict the effect of a new combination of drugs without having to test it experimentally can help in accelerating this process. In particular, network-based dynamic models of these pathways hold promise to both understand and predict the effect of therapeutics. However, their use is currently hampered by limitations in our knowledge of the underlying biochemistry, as well as in the experimental and computational technologies used for calibrating the models. Thus, the results from such models need to be carefully interpreted and used in order to avoid biased predictions. Here we present a procedure that deals with this uncertainty by using experimental data to build an ensemble of dynamic models. The method incorporates steps to reduce overfitting and maximize predictive capability. We find that by combining the outputs of individual models in an ensemble it is possible to obtain a more robust prediction. We report results obtained with this method, which we call SELDOM (enSEmbLe of Dynamic lOgic-based Models), showing that it improves the predictions previously reported for several challenging problems. ; JRB and DH acknowledge funding from the EU FP7 project NICHE (ITN Grant number 289384). JRB acknowledges funding from the Spanish MINECO project SYNBIOFACTORY (grant number DPI2014-55276-C5-2-R). AFV acknowledges funding from the Galician government (Xunta de Galiza) through the I2C postdoctoral fellowship ED481B2014/133-0. The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript. ...