The importance of graph databases and graph learning for clinical applications.

The increasing amount and complexity of clinical data require an appropriate way of storing and analyzing those data. Traditional approaches use a tabular structure (relational databases) for storing data and thereby complicate storing and retrieving interlinked data from the clinical domain. Graph databases provide a great solution for this by storing data in a graph as nodes (vertices) that are connected by edges (links). The underlying graph structure can be used for the subsequent data analysis (graph learning). Graph learning consists of two parts: graph representation learning and graph analytics. Graph representation learning aims to reduce high-dimensional input graphs to low-dimensional representations. Then, graph analytics uses the obtained representations for analytical tasks like visualization, classification, link prediction and clustering which can be used to solve domain-specific problems. In this survey, we review current state-of-the-art graph database management systems, graph learning algorithms and a variety of graph applications in the clinical domain. Furthermore, we provide a comprehensive use case for a clearer understanding of complex graph learning algorithms. Graphical abstract.

Decision tree learning in Neo4j on homogeneous and unconnected graph nodes from biological and clinical datasets.

BACKGROUND: Graph databases enable efficient storage of heterogeneous, highly-interlinked data, such as clinical data. Subsequently, researchers can extract relevant features from these datasets and apply machine learning for diagnosis, biomarker discovery, or understanding pathogenesis. METHODS: To facilitate machine learning and save time for extracting data from the graph database, we developed and optimized Decision Tree Plug-in (DTP) containing 24 procedures to generate and evaluate decision trees directly in the graph database Neo4j on homogeneous and unconnected nodes. RESULTS: Creation of the decision tree for three clinical datasets directly in the graph database from the nodes required between 0.059 and 0.099 s, while calculating the decision tree with the same algorithm in Java from CSV files took 0.085-0.112 s. Furthermore, our approach was faster than the standard decision tree implementations in R (0.62 s) and equal to Python (0.08 s), also using CSV files as input for small datasets. In addition, we have explored the strengths of DTP by evaluating a large dataset (approx. 250,000 instances) to predict patients with diabetes and compared the performance against algorithms generated by state-of-the-art packages in R and Python. By doing so, we have been able to show competitive results on the performance of Neo4j, in terms of quality of predictions as well as time efficiency. Furthermore, we could show that high body-mass index and high blood pressure are the main risk factors for diabetes. CONCLUSION: Overall, our work shows that integrating machine learning into graph databases saves time for additional processes as well as external memory, and could be applied to a variety of use cases, including clinical applications. This provides user with the advantages of high scalability, visualization and complex querying.

Fecal Metaproteomics Reveals Reduced Gut Inflammation and Changed Microbial Metabolism Following Lifestyle-Induced Weight Loss.

Gut microbiota-mediated inflammation promotes obesity-associated low-grade inflammation, which represents a hallmark of metabolic syndrome. To investigate if lifestyle-induced weight loss (WL) may modulate the gut microbiome composition and its interaction with the host on a functional level, we analyzed the fecal metaproteome of 33 individuals with metabolic syndrome in a longitudinal study before and after lifestyle-induced WL in a well-defined cohort. The 6-month WL intervention resulted in reduced BMI (-13.7%), improved insulin sensitivity (HOMA-IR, -46.1%), and reduced levels of circulating hsCRP (-39.9%), indicating metabolic syndrome reversal. The metaprotein spectra revealed a decrease of human proteins associated with gut inflammation. Taxonomic analysis revealed only minor changes in the bacterial composition with an increase of the families Desulfovibrionaceae, Leptospiraceae, Syntrophomonadaceae, Thermotogaceae and Verrucomicrobiaceae. Yet we detected an increased abundance of microbial metaprotein spectra that suggest an enhanced hydrolysis of complex carbohydrates. Hence, lifestyle-induced WL was associated with reduced gut inflammation and functional changes of human and microbial enzymes for carbohydrate hydrolysis while the taxonomic composition of the gut microbiome remained almost stable. The metaproteomics workflow has proven to be a suitable method for monitoring inflammatory changes in the fecal metaproteome.

A Robust and Universal Metaproteomics Workflow for Research Studies and Routine Diagnostics Within 24 h Using Phenol Extraction, FASP Digest, and the MetaProteomeAnalyzer.

The investigation of microbial proteins by mass spectrometry (metaproteomics) is a key technology for simultaneously assessing the taxonomic composition and the functionality of microbial communities in medical, environmental, and biotechnological applications. We present an improved metaproteomics workflow using an updated sample preparation and a new version of the MetaProteomeAnalyzer software for data analysis. High resolution by multidimensional separation (GeLC, MudPIT) was sacrificed to aim at fast analysis of a broad range of different samples in less than 24 h. The improved workflow generated at least two times as many protein identifications than our previous workflow, and a drastic increase of taxonomic and functional annotations. Improvements of all aspects of the workflow, particularly the speed, are first steps toward potential routine clinical diagnostics (i.e., fecal samples) and analysis of technical and environmental samples. The MetaProteomeAnalyzer is provided to the scientific community as a central remote server solution at www.mpa.ovgu.de.

Challenges and perspectives of metaproteomic data analysis.

In nature microorganisms live in complex microbial communities. Comprehensive taxonomic and functional knowledge about microbial communities supports medical and technical application such as fecal diagnostics as well as operation of biogas plants or waste water treatment plants. Furthermore, microbial communities are crucial for the global carbon and nitrogen cycle in soil and in the ocean. Among the methods available for investigation of microbial communities, metaproteomics can approximate the activity of microorganisms by investigating the protein content of a sample. Although metaproteomics is a very powerful method, issues within the bioinformatic evaluation impede its success. In particular, construction of databases for protein identification, grouping of redundant proteins as well as taxonomic and functional annotation pose big challenges. Furthermore, growing amounts of data within a metaproteomics study require dedicated algorithms and software. This review summarizes recent metaproteomics software and addresses the introduced issues in detail.