Usefulness of Heat Map Explanations for Deep-Learning-Based Electrocardiogram Analysis.

Deep neural networks are complex machine learning models that have shown promising results in analyzing high-dimensional data such as those collected from medical examinations. Such models have the potential to provide fast and accurate medical diagnoses. However, the high complexity makes deep neural networks and their predictions difficult to understand. Providing model explanations can be a way of increasing the understanding of "black box" models and building trust. In this work, we applied transfer learning to develop a deep neural network to predict sex from electrocardiograms. Using the visual explanation method Grad-CAM, heat maps were generated from the model in order to understand how it makes predictions. To evaluate the usefulness of the heat maps and determine if the heat maps identified electrocardiogram features that could be recognized to discriminate sex, medical doctors provided feedback. Based on the feedback, we concluded that, in our setting, this mode of explainable artificial intelligence does not provide meaningful information to medical doctors and is not useful in the clinic. Our results indicate that improved explanation techniques that are tailored to medical data should be developed before deep neural networks can be applied in the clinic for diagnostic purposes.

Endosomal egress and intercellular transmission of hepatic ApoE-containing lipoproteins and its exploitation by the hepatitis C virus.

Liver-generated plasma Apolipoprotein E (ApoE)-containing lipoproteins (LPs) (ApoE-LPs) play central roles in lipid transport and metabolism. Perturbations of ApoE can result in several metabolic disorders and ApoE genotypes have been associated with multiple diseases. ApoE is synthesized at the endoplasmic reticulum and transported to the Golgi apparatus for LP assembly; however, the ApoE-LPs transport pathway from there to the plasma membrane is largely unknown. Here, we established an integrative imaging approach based on a fully functional fluorescently tagged ApoE. We found that newly synthesized ApoE-LPs accumulate in CD63-positive endosomes of hepatocytes. In addition, we observed the co-egress of ApoE-LPs and CD63-positive intraluminal vesicles (ILVs), which are precursors of extracellular vesicles (EVs), along the late endosomal trafficking route in a microtubule-dependent manner. A fraction of ApoE-LPs associated with CD63-positive EVs appears to be co-transmitted from cell to cell. Given the important role of ApoE in viral infections, we employed as well-studied model the hepatitis C virus (HCV) and found that the viral replicase component nonstructural protein 5A (NS5A) is enriched in ApoE-containing ILVs. Interaction between NS5A and ApoE is required for the efficient release of ILVs containing HCV RNA. These vesicles are transported along the endosomal ApoE egress pathway. Taken together, our data argue for endosomal egress and transmission of hepatic ApoE-LPs, a pathway that is hijacked by HCV. Given the more general role of EV-mediated cell-to-cell communication, these insights provide new starting points for research into the pathophysiology of ApoE-related metabolic and infection-related disorders.

Drivers and determinants of strain dynamics following fecal microbiota transplantation.

Fecal microbiota transplantation (FMT) is a therapeutic intervention for inflammatory diseases of the gastrointestinal tract, but its clinical mode of action and subsequent microbiome dynamics remain poorly understood. Here we analyzed metagenomes from 316 FMTs, sampled pre and post intervention, for the treatment of ten different disease indications. We quantified strain-level dynamics of 1,089 microbial species, complemented by 47,548 newly constructed metagenome-assembled genomes. Donor strain colonization and recipient strain resilience were mostly independent of clinical outcomes, but accurately predictable using LASSO-regularized regression models that accounted for host, microbiome and procedural variables. Recipient factors and donor-recipient complementarity, encompassing entire microbial communities to individual strains, were the main determinants of strain population dynamics, providing insights into the underlying processes that shape the post-FMT gut microbiome. Applying an ecology-based framework to our findings indicated parameters that may inform the development of more effective, targeted microbiome therapies in the future, and suggested how patient stratification can be used to enhance donor microbiota colonization or the displacement of recipient microbes in clinical practice.

metaSNV v2: detection of SNVs and subspecies in prokaryotic metagenomes.

SUMMARY: Taxonomic analysis of microbial communities is well supported at the level of species and strains. However, species can contain significant phenotypic diversity and strains are rarely widely shared across global populations. Stratifying the diversity between species and strains can identify 'subspecies', which are a useful intermediary. High-throughput identification and profiling of subspecies is not yet supported in the microbiome field. Here, we use an operational definition of subspecies based on single nucleotide variant (SNV) patterns within species to identify and profile subspecies in metagenomes, along with their distinctive SNVs and genes. We incorporate this method into metaSNV v2, which extends existing SNV-calling software to support further SNV interpretation for population genetics. These new features support microbiome analyses to link SNV profiles with host phenotype or environment and niche-specificity. We demonstrate subspecies identification in marine and fecal metagenomes. In the latter, we analyze 70 species in 7524 adult and infant subjects, supporting a common subspecies population structure in the human gut microbiome and illustrating some limits in subspecies calling. AVAILABILITY AND IMPLEMENTATION: Source code, documentation, tutorials and test data are available at https://github.com/metasnv-tool/metaSNV and https://metasnv.embl.de. SUPPLEMENTARY INFORMATION: Supplementary data are available at Bioinformatics online.