Development of a new data management system for the study of the gut microbiome of children who are small for their gestational age

Abstract

Microbiome studies aim to answer the following questions: which organisms are in the sample and what is their impact on the patient or the environment? To answer these questions, investigators have to perform comparative analyses on their classified sequences based on the collected metadata, such as treatment, condition of the patient, or the environment. The integrity of sequences, classifications, and metadata is paramount for the success of such studies. Still, the area of data management for the preliminary study results appears to be neglected. Here, we present the development of MetagenomicsDB (http://github.com/IOB-Muenster/MetagenomicsDB; accessed 2024/12/18), a central data management system for the study of the gut microbiome in children who are small for their gestational age (SGA). Our system provided more flexibility to conduct study-specific analyses and to integrate specific external resources than existing and necessarily more generic solutions. It supports short or long read data produced by virtually any sequencing instrument targeting (parts of) popular marker genes, such as the 16S rRNA gene and its variable regions. Classifications of these reads from the MetaG and Kraken 2 software are supported. The main goals of the system are to store the pre-computed study data securely under concurrent load and to make downstream analyses accessible to all researchers, regardless of programming proficiency. Thus, after initial plausibility checks on the input data to reduce human error, data are stored in a relational database and can be continuously updated over the whole life time of the study. We used a modular approach for MetagenomicsDB with comprehensive tests verifying the expected behavior and extensively described the underlying rational which allows users to adapt the system to their needs. We advocate the use of MetagenomicsDB as the backend for a graphical web interface. We showcase the potential of this approach at the example of our study on SGA children (http://www.bioinformatics.uni-muenster.de/tools/metagenomicsDB; accessed 2024/12/02). Without restrictions caused by the level of programming proficiency, our team members could explore the study data and optionally filter them using the graphical interface, before exporting the data in a format directly suitable for external normalization of read counts and statistical analyses. Study results could be conveniently and transparently shared with the public, as demonstrated here. Links to external resources facilitated literature search with regard to the SGA condition and assessments of the potential pathogenicity of taxa. Since different users will have different demands regarding features, data security, and web environments, we provide our implementation of the web interface as a visual example. By providing users with the MetagenomicsDB backend which constitutes the major part of the system, we ensure that custom development can be finished in a reasonable amount of time. We report our endeavors in order to motivate the application of data management systems at the scale of single studies in microbiome research.