RESULTS:
1 - 2 of 2 for "Edi Prifti"
Impact of simulation and reference catalogues on the evaluation of taxonomic profiling pipelines
Microbiome profiling tools rely on reference catalogues which significantly affect their performance. Comparing them is however challenging mainly due to differences in their native catalogues. In this study we present a novel standardized benchmarking framework that makes such comparisons more accurate. We decided not to customize databases but to translate results to a common reference to use the tools with their native environment. Specifically we conducted two realistic simulations of gut microbiome samples each based on a specific taxonomic profiler and used two different taxonomic references to project their results namely the Genome Taxonomy Database and the Unified Human Gastrointestinal Genome. To demonstrate the importance of using such a framework we evaluated four established profilers as well as the impact of the simulations and that of the common taxonomic references on the perceived performance of these profilers. Finally we provide guidelines to enhance future profiler comparisons for human microbiome ecosystems: (i) use or create realistic simulations tailored to your biological context (BC) (ii) identify a common feature space suited to your BC and independent of the catalogues used by the profilers and (iii) apply a comprehensive set of metrics covering accuracy (sensitivity/precision) overall representativity (richness/Shannon) and quantification (UniFrac and/or Aitchison distance).
Deep learning methods in metagenomics: a review
The ever-decreasing cost of sequencing and the growing potential applications of metagenomics have led to an unprecedented surge in data generation. One of the most prevalent applications of metagenomics is the study of microbial environments such as the human gut. The gut microbiome plays a crucial role in human health providing vital information for patient diagnosis and prognosis. However analysing metagenomic data remains challenging due to several factors including reference catalogues sparsity and compositionality. Deep learning (DL) enables novel and promising approaches that complement state-of-the-art microbiome pipelines. DL-based methods can address almost all aspects of microbiome analysis including novel pathogen detection sequence classification patient stratification and disease prediction. Beyond generating predictive models a key aspect of these methods is also their interpretability. This article reviews DL approaches in metagenomics including convolutional networks autoencoders and attention-based models. These methods aggregate contextualized data and pave the way for improved patient care and a better understanding of the microbiome’s key role in our health.