Structure-revealing data fusion

BACKGROUND

Analysis of data from multiple sources has the potential to enhance knowledge discovery by capturing underlying structures, which are, otherwise, difficult to extract. Fusing data from multiple sources has already proved useful in many applications in social network analysis, signal processing and bioinformatics. However, data fusion is challenging since data from multiple sources are often (i) heterogeneous (i.e., in the form of higher-order tensors and matrices), (ii) incomplete, and (iii) have both shared and unshared components. In order to address these challenges, in this paper, we introduce a novel unsupervised data fusion model based on joint factorization of matrices and higher-order tensors.

RESULTS

While the traditional formulation of coupled matrix and tensor factorizations modeling only shared factors fails to capture the underlying structures in the presence of both shared and unshared factors, the proposed data fusion model has the potential to automatically reveal shared and unshared components through modeling constraints. Using numerical experiments, we demonstrate the effectiveness of the proposed approach in terms of identifying shared and unshared components. Furthermore, we measure a set of mixtures with known chemical composition using both LC-MS (Liquid Chromatography - Mass Spectrometry) and NMR (Nuclear Magnetic Resonance) and demonstrate that the structure-revealing data fusion model can (i) successfully capture the chemicals in the mixtures and extract the relative concentrations of the chemicals accurately, (ii) provide promising results in terms of identifying shared and unshared chemicals, and (iii) reveal the relevant patterns in LC-MS by coupling with the diffusion NMR data.

CONCLUSIONS

We have proposed a structure-revealing data fusion model that can jointly analyze heterogeneous, incomplete data sets with shared and unshared components and demonstrated its promising performance as well as potential limitations on both simulated and real data.

The food metabolome: a window over dietary exposure

The food metabolome is defined as the part of the human metabolome directly derived from the digestion and biotransformation of foods and their constituents. With >25,000 compounds known in various foods, the food metabolome is extremely complex, with a composition varying widely according to the diet. By its very nature it represents a considerable and still largely unexploited source of novel dietary biomarkers that could be used to measure dietary exposures with a high level of detail and precision. Most dietary biomarkers currently have been identified on the basis of our knowledge of food compositions by using hypothesis-driven approaches. However, the rapid development of metabolomics resulting from the development of highly sensitive modern analytic instruments, the availability of metabolite databases, and progress in (bio)informatics has made agnostic approaches more attractive as shown by the recent identification of novel biomarkers of intakes for fruit, vegetables, beverages, meats, or complex diets. Moreover, examples also show how the scrutiny of the food metabolome can lead to the discovery of bioactive molecules and dietary factors associated with diseases. However, researchers still face hurdles, which slow progress and need to be resolved to bring this emerging field of research to maturity. These limits were discussed during the First International Workshop on the Food Metabolome held in Glasgow. Key recommendations made during the workshop included more coordination of efforts; development of new databases, software tools, and chemical libraries for the food metabolome; and shared repositories of metabolomic data. Once achieved, major progress can be expected toward a better understanding of the complex interactions between diet and human health.