Refine
Year of publication
Document Type
- Article (63) (remove)
Has Fulltext
- no (63)
Is part of the Bibliography
- no (63)
Keywords
- ANFIS (1)
- ANN (1)
- Cross-validation (1)
- Disease severity (1)
- Epidemiology (1)
- Infectious Diseases (1)
- MODIS (1)
- Microbiology (1)
- Parasitology (1)
- Public Health, Environmental and Occupational Health (1)
Analyzing mass spectrometry-based proteomics data with deep learning (DL) approaches poses several challenges due to the high dimensionality, low sample size, and high level of noise. Additionally, DL-based workflows are often hindered to be integrated into medical settings due to the lack of interpretable explanation. We present DLearnMS, a DL biomarker detection framework, to address these challenges on proteomics instances of liquid chromatography-mass spectrometry (LC-MS) - a well-established tool for quantifying complex protein mixtures. Our DLearnMS framework learns the clinical state of LC-MS data instances using convolutional neural networks. Based on the trained neural networks, we show how biomarkers can be identified using layer-wise relevance propagation. This enables detecting discriminating regions of the data and the design of more robust networks. One of the main advantages over other established methods is that no explicit preprocessing step is needed in our DLearnMS framework. Our evaluation shows that DLearnMS outperforms conventional LC-MS biomarker detection approaches in identifying fewer false positive peaks while maintaining a comparable amount of true positives peaks.
Large-scale perturbations in the microbiome constitution are strongly correlated, whether as a driver or a consequence, with the health and functioning of human physiology. However, understanding the difference in the microbiome profiles of healthy and ill individuals can be complicated due to the large number of complex interactions among microbes. We propose to model these interactions as a time-evolving graph whose nodes are microbes and edges are interactions among them. Motivated by the need to analyse such complex interactions, we develop a method that learns a low-dimensional representation of the time-evolving graph and maintains the dynamics occurring in the high-dimensional space. Through our experiments, we show that we can extract graph features such as clusters of nodes or edges that have the highest impact on the model to learn the low-dimensional representation. This information can be crucial to identify microbes and interactions among them that are strongly correlated with clinical diseases. We conduct our experiments on both synthetic and real-world microbiome datasets.
Various feature selection algorithms have been proposed to identify cancer prognostic biomarkers. In recent years, however, their reproducibility is criticized. The performance of feature selection algorithms is shown to be affected by the datasets, underlying networks and evaluation metrics. One of the causes is the curse of dimensionality, which makes it hard to select the features that generalize well on independent data. Even the integration of biological networks does not mitigate this issue because the networks are large and many of their components are not relevant for the phenotype of interest. With the availability of multi-omics data, integrative approaches are being developed to build more robust predictive models. In this scenario, the higher data dimensions create greater challenges. We proposed a phenotype relevant network-based feature selection (PRNFS) framework and demonstrated its advantages in lung cancer prognosis prediction. We constructed cancer prognosis relevant networks based on epithelial mesenchymal transition (EMT) and integrated them with different types of omics data for feature selection. With less than 2.5% of the total dimensionality, we obtained EMT prognostic signatures that achieved remarkable prediction performance (average AUC values above 0.8), very significant sample stratifications, and meaningful biological interpretations. In addition to finding EMT signatures from different omics data levels, we combined these single-omics signatures into multi-omics signatures, which improved sample stratifications significantly. Both single- and multi-omics EMT signatures were tested on independent multi-omics lung cancer datasets and significant sample stratifications were obtained.
Motivation: High-throughput proteomics techniques, such as mass spectrometry
(MS)-based approaches, produce very high-dimensional data-sets. In a clinical
setting one is often interested how MS spectra dier between patients of different
classes, for example spectra from healthy patients vs. spectra from patients
having a particular disease. Machine learning algorithms are needed to (a)
identify these discriminating features and (b) classify unknown spectra based on
this feature set. Since the acquired data is usually noisy, the algorithms should be
robust to noise and outliers, and the identied feature set should be as small as
possible.
Results: We present a new algorithm, Sparse Proteomics Analysis (SPA), based
on the theory of Compressed Sensing that allows to identify a minimal
discriminating set of features from mass spectrometry data-sets. We show how
our method performs on artificial and real-world data-sets.
Epithelial mesenchymal transition (EMT) process has been shown as highly relevant to cancer prognosis. However, although different biological network-based biomarker identification methods have been proposed to predict cancer prognosis, EMT network has not been directly used for this purpose. In this study, we constructed an EMT regulatory network consisting of 87 molecules and tried to select features that are useful for prognosis prediction in Lung Adenocarcinoma (LUAD). To incorporate multiple molecular profiles, we obtained four types of molecular data including mRNA-Seq, copy number alteration (CNA), DNA methylation, and miRNA-Seq data from The Cancer Genome Atlas. The data were mapped to the EMT network in three alternative ways: mRNA-Seq and miRNA-Seq, DNA methylation, and CNA and miRNA-Seq. Each mapping was employed to extract five different sets of features using discretization and network-based biomarker identification methods. Each feature set was then used to predict prognosis with SVM and logistic regression classifiers. We measured the prediction accuracy with AUC and AUPR values using 10 times 10-fold cross validation. For a more comprehensive evaluation, we also measured the prediction accuracies of clinical features, EMT plus clinical features, randomly picked 87 molecules from each data mapping, and using all molecules from each data type. Counter-intuitively, EMT features do not always outperform randomly selected features and the prediction accuracies of the five feature sets are mostly not significantly different. Clinical features are shown to give the highest prediction accuracies. In addition, the prediction accuracies of both EMT features and random features are comparable as using all features (more than 17,000) from each data type.
Mass spectrometry-based serum metabolic profiling is a promising tool to analyse complex cancer associated metabolic alterations, which may broaden our pathophysiological understanding of the disease and may function as a source of new cancer-associated biomarkers. Highly standardized serum samples of patients suffering from colon cancer (n = 59) and controls (n = 58) were collected at the University Hospital Leipzig. We based our investigations on amino acid screening profiles using electrospray tandem-mass spectrometry. Metabolic profiles were evaluated using the Analyst 1.4.2 software. General, comparative and equivalence statistics were performed by R 2.12.2. 11 out of 26 serum amino acid concentrations were significantly different between colorectal cancer patients and healthy controls. We found a model including CEA, glycine, and tyrosine as best discriminating and superior to CEA alone with an AUROC of 0.878 (95\% CI 0.815?0.941). Our serum metabolic profiling in colon cancer revealed multiple significant disease-associated alterations in the amino acid profile with promising diagnostic power. Further large-scale studies are necessary to elucidate the potential of our model also to discriminate between cancer and potential differential diagnoses. In conclusion, serum glycine and tyrosine in combination with CEA are superior to CEA for the discrimination between colorectal cancer patients and controls.
Background: High-throughput proteomics techniques, such as mass spectrometry (MS)-based approaches, produce very high-dimensional data-sets. In a clinical setting one is often interested in how mass spectra differ between patients of different classes, for example spectra from healthy patients vs. spectra from patients having a particular disease. Machine learning algorithms are needed to (a) identify these discriminating features and (b) classify unknown spectra based on this feature set. Since the acquired data is usually noisy, the algorithms should be robust against noise and outliers, while the identified feature set should be as small as possible.
Results: We present a new algorithm, Sparse Proteomics Analysis (SPA),based on thet heory of compressed sensing that allows us to identify a minimal discriminating set of features from mass spectrometry data-sets. We show (1) how our method performs on artificial and real-world data-sets, (2) that its performance is competitive with standard (and widely used) algorithms for analyzing proteomics data, and (3) that it is robust against random and systematic noise. We further demonstrate the applicability of our algorithm to two previously published clinical data-sets.
In the framework of time series analysis with recurrence networks, we introduce a self-adaptive method that determines the elusive recurrence threshold and identifies metastable states in complex real-world time series. As initial step, we introduce a way to set the embedding parameters used to reconstruct the state space from the time series. We set them as the ones giving the maximum Shannon entropy of the diagonal line length distribution for the first simultaneous minima of recurrence rate and Shannon entropy. To identify metastable states, as well as the transitions between them, we use a soft partitioning algorithm for module finding which is specifically developed for the case in which a system shows metastability. We illustrate our method with a complex time series example. Finally, we show the robustness of our method for identifying metastable states. Our results suggest that our method is robust for identifying metastable states in complex time series, even when introducing considerable levels of noise and missing data points.