Model-Based Explainable Deep Learning for Light-Field Microscopy Imaging

In modern neuroscience, observing the dynamics of large populations of neurons is a critical step of understanding how networks of neurons process information. Light-field microscopy (LFM) has emerged as a type of scanless, high-speed, three-dimensional (3D) imaging tool, particularly attractive for this purpose. Imaging neuronal activity using LFM calls for the development of novel computational approaches that fully exploit domain knowledge embedded in physics and optics models, as well as enabling high interpretability and transparency. To this end, we propose a model-based explainable deep learning approach for LFM. Different from purely data-driven methods, the proposed approach integrates wave-optics theory, sparse representation and non-linear optimization with the artificial neural network. In particular, the architecture of the proposed neural network is designed following precise signal and optimization models. Moreover, the network's parameters are learned from a training dataset using a novel training strategy that integrates layer-wise training with tailored knowledge distillation. Such design allows the network to take advantage of domain knowledge and learned new features. It combines the benefit of both model-based and learning-based methods, thereby contributing to superior interpretability, transparency and performance. By evaluating on both structural and functional LFM data obtained from scattering mammalian brain tissues, we demonstrate the capabilities of the proposed approach to achieve fast, robust 3D localization of neuron sources and accurate neural activity identification.

Population genetics and microevolution of clinical Candida glabrata reveals recombinant sequence types and hyper-variation within mitochondrial genomes, virulence genes, and drug targets

Candida glabrata is the second most common etiological cause of worldwide systemic candidiasis in adult patients. Genome analysis of 68 isolates from 8 hospitals across Scotland, together with 83 global isolates, revealed insights into the population genetics and evolution of C. glabrata. Clinical isolates of C. glabrata from across Scotland are highly genetically diverse, including at least 19 separate sequence types that have been recovered previously in globally diverse locations, and 1 newly discovered sequence type. Several sequence types had evidence for ancestral recombination, suggesting transmission between distinct geographical regions has coincided with genetic exchange arising in new clades. Three isolates were missing MATα1, potentially representing a second mating type. Signatures of positive selection were identified in every sequence type including enrichment for epithelial adhesins thought to facilitate fungal adhesin to human epithelial cells. In patent microevolution was identified from 7 sets of recurrent cases of candidiasis, revealing an enrichment for nonsynonymous and frameshift indels in cell surface proteins. Microevolution within patients also affected epithelial adhesins genes, and several genes involved in drug resistance including the ergosterol synthesis gene ERG4 and the echinocandin target FKS1/2, the latter coinciding with a marked drop in fluconazole minimum inhibitory concentration. In addition to nuclear genome diversity, the C. glabrata mitochondrial genome was particularly diverse, with reduced conserved sequence and conserved protein-encoding genes in all nonreference ST15 isolates. Together, this study highlights the genetic diversity within the C. glabrata population that may impact virulence and drug resistance, and 2 major mechanisms generating this diversity: microevolution and genetic exchange/recombination.

Drivers of methicillin-resistant Staphylococcus aureus (MRSA) lineage replacement in China

BACKGROUND

Methicillin-resistant Staphylococcus aureus (MRSA) is a major nosocomial pathogen subdivided into lineages termed sequence types (STs). Since the 1950s, successive waves of STs have appeared and replaced previously dominant lineages. One such event has been occurring in China since 2013, with community-associated (CA-MRSA) strains including ST59 largely replacing the previously dominant healthcare-associated (HA-MRSA) ST239. We previously showed that ST59 isolates tend to have a competitive advantage in growth experiments against ST239. However, the underlying genomic and phenotypic drivers of this replacement event are unclear.

METHODS

Here, we investigated the replacement of ST239 using whole-genome sequencing data from 204 ST239 and ST59 isolates collected in Chinese hospitals between 1994 and 2016. We reconstructed the evolutionary history of each ST and considered two non-mutually exclusive hypotheses for ST59 replacing ST239: antimicrobial resistance (AMR) profile and/or ability to colonise and persist in the environment through biofilm formation. We also investigated the differences in cytolytic activity, linked to higher virulence, between STs. We performed an association study using the presence and absence of accessory virulence genes.

RESULTS

ST59 isolates carried fewer AMR genes than ST239 and showed no evidence of evolving towards higher AMR. Biofilm production was marginally higher in ST59 overall, though this effect was not consistent across sub-lineages so is unlikely to be a sole driver of replacement. Consistent with previous observations of higher virulence in CA-MRSA STs, we observed that ST59 isolates exhibit significantly higher cytolytic activity than ST239 isolates, despite carrying on average fewer putative virulence genes. Our association study identified the chemotaxis inhibitory protein (chp) as a strong candidate for involvement in the increased virulence potential of ST59. We experimentally validated the role of chp in increasing the virulence potential of ST59 by creating Δchp knockout mutants, confirming that ST59 can carry chp without a measurable impact on fitness.

CONCLUSIONS

Our results suggest that the ongoing replacement of ST239 by ST59 in China is not associated to higher AMR carriage or biofilm production. However, the increase in ST59 prevalence is concerning since it is linked to a higher potential for virulence, aided by the carriage of the chp gene.

Efficient communication over complex dynamical networks: The role of matrix non-normality

In both natural and engineered systems, communication often occurs dynamically over networks ranging from highly structured grids to largely disordered graphs. To use, or comprehend the use of, networks as efficient communication media requires understanding of how they propagate and transform information in the face of noise. Here, we develop a framework that enables us to examine how network structure, noise, and interference between consecutive packets jointly determine transmission performance in complex networks governed by linear dynamics. Mathematically, normal networks, which can be decomposed into separate low-dimensional information channels, suffer greatly from readout noise. Most details of their wiring have no impact on transmission quality. Non-normal networks, however, can largely cancel the effect of noise by transiently amplifying select input dimensions while ignoring others, resulting in higher net information throughput. Our theory could inform the design of new communication networks, as well as the optimal use of existing ones.

Robust Phagocyte Recruitment Controls the Opportunistic Fungal Pathogen Mucor circinelloides in Innate Granulomas In Vivo

Mucormycosis is an emerging fungal infection with extremely high mortality rates in patients with defects in their innate immune response, specifically in functions mediated through phagocytes. However, we currently have a limited understanding of the molecular and cellular interactions between these innate immune effectors and mucormycete spores during the early immune response. Here, the early events of innate immune recruitment in response to infection by Mucor circinelloides spores are modeled by a combined in silico modeling approach and real-time in vivo microscopy. Phagocytes are rapidly recruited to the site of infection in a zebrafish larval model of mucormycosis. This robust early recruitment protects from disease onset in vivoIn silico analysis identified that protection is dependent on the number of phagocytes at the infection site, but not the speed of recruitment. The mathematical model highlights the role of proinflammatory signals for phagocyte recruitment and the importance of inhibition of spore germination for protection from active fungal disease. These in silico data are supported by an in vivo lack of fungal spore killing and lack of reactive oxygen burst, which together result in latent fungal infection. During this latent stage of infection, spores are controlled in innate granulomas in vivo Disease can be reactivated by immunosuppression. Together, these data represent the first in vivo real-time analysis of innate granuloma formation during the early stages of a fungal infection. The results highlight a potential latent stage during mucormycosis that should urgently be considered for clinical management of patients.IMPORTANCE Mucormycosis is a dramatic fungal infection frequently leading to the death of patients. We know little about the immune response to the fungus causing this infection, although evidence points toward defects in early immune events after infection. Here, we dissect this early immune response to infectious fungal spores. We show that specialized white blood cells (phagocytes) rapidly respond to these spores and accumulate around the fungus. However, we demonstrate that the mechanisms that enable phagocytes to kill the fungus fail, allowing for survival of spores. Instead a cluster of phagocytes resembling an early granuloma is formed around spores to control the latent infection. This study is the first detailed analysis of early granuloma formation during a fungal infection highlighting a latent stage that needs to be considered for clinical management of patients.