Logo: to the web site of Uppsala University

Electroencephalogram (EEG) interpretation plays a critical role in the clinical assessment of neurological conditions, most notably epilepsy. However, EEG recordings are typically analyzed manually by highly specialized and heavily trained personnel. Moreover, the low rate of capturing abnormal events during the procedure makes interpretation time-consuming, resource-hungry, and overall an expensive process. Automatic detection offers the potential to improve the quality of patient care by shortening the time to diagnosis, managing big data and optimizing the allocation of human resources towards precision medicine. Here, we present MindReader, a novel unsupervised machine-learning method comprised of the interplay between an autoencoder network, a hidden Markov model (HMM), and a generative component: after dividing the signal into overlapping frames and performing a fast Fourier transform, MindReader trains an autoencoder neural network for dimensionality reduction and compact representation of different frequency patterns for each frame. Next, we processed the temporal patterns using a HMM, while a third and generative component hypothesized and characterized the different phases that were then fed back to the HMM. MindReader then automatically generates labels that the physician can interpret as pathological and non-pathological phases, thus effectively reducing the search space for trained personnel. We evaluated MindReader’s predictive performance on 686 recordings, encompassing more than 980 h from the publicly available Physionet database. Compared to manual annotations, MindReader identified 197 of 198 epileptic events (99.45%), and is, as such, a highly sensitive method, which is a prerequisite for clinical use.

Background: Electroencephalogram (EEG) interpretation plays a critical role in the clinical assessment of neurological conditions, including epilepsy. Manual analysis requires highly specialized and heavily trained personnel. Moreover, the rate of capturing abnormal events makes interpretation time-consuming, resource-hungry, and, overall, an expensive process.

Automatic detection offers the potential to improve the quality of patient care by shortening the time to diagnosis, managing big data, and optimizing the allocation of human resources.

Findings: We present MindReader, an unsupervised method for EEG signals. First, MindReader processes the signal through an autoencoder in order to detect EEG abnormalities. Next, patterns are hypothesized by a Hidden Markov Model. Our algorithm automatically generates labels for non-pathological phases, thus reducing the search space for trained personnel.

Conclusions: MindReader is effective in detecting EEG abnormalities in focal and generalized epilepsy.

The current gold standard of gait diagnostics is dependent on large, expensive motion-capture laboratories and highly trained clinical and technical staff. Wearable sensor systems combined with machine learning may help to improve the accessibility of objective gait assessments in a broad clinical context. However, current algorithms lack flexibility and require large training datasets with tedious manual labelling of data. The current study tests the validity of a novel machine learning algorithm for automated gait partitioning of laboratory-based and sensor-based gait data. The developed artificial intelligence tool was used in patients with a central neurological lesion and severe gait impairments. To build the novel algorithm, 2% and 3% of the entire dataset (567 and 368 steps in total, respectively) were required for assessments with laboratory equipment and inertial measurement units. The mean errors of machine learning-based gait partitions were 0.021 s for the laboratory-based datasets and 0.034 s for the sensor-based datasets. Combining reinforcement learning with a deep neural network allows significant reduction in the size of the training datasets to <5%. The low number of required training data provides end-users with a high degree of flexibility. Non-experts can easily adjust the developed algorithm and modify the training library depending on the measurement system and clinical population.

Carbon storage and cycling in boreal forests-the largest terrestrial carbon store-is moderated by complex interactions between trees and soil microorganisms. However, existing methods limit our ability to predict how changes in environmental conditions will alter these associations and the essential ecosystem services they provide. To address this, we developed a metatranscriptomic approach to analyze the impact of nutrient enrichment on Norway spruce fine roots and the community structure, function, and tree-microbe coordination of over 350 root-associated fungal species. In response to altered nutrient status, host trees redefined their relationship with the fungal community by reducing sugar efflux carriers and enhancing defense processes. This resulted in a profound restructuring of the fungal community and a collapse in functional coordination between the tree and the dominant Basidiomycete species, and an increase in functional coordination with versatile Ascomycete species. As such, there was a functional shift in community dominance from Basidiomycetes species, with important roles in enzymatically cycling recalcitrant carbon, to Ascomycete species that have melanized cell walls that are highly resistant to degradation. These changes were accompanied by prominent shifts in transcriptional coordination between over 60 predicted fungal effectors, with more than 5,000 Norway spruce transcripts, providing mechanistic insight into the complex molecular dialogue coordinating host trees and their fungal partners. The host-microbe dynamics captured by this study functionally inform how these complex and sensitive biological relationships may mediate the carbon storage potential of boreal soils under changing nutrient conditions.

The health, growth, and fitness of boreal forest trees are impacted and improved by their associated microbiomes. Microbial gene expression and functional activity can be assayed with RNA sequencing (RNA-Seq) data from host samples. In contrast, phylogenetic marker gene amplicon sequencing data are used to assess taxonomic composition and community structure of the microbiome. Few studies have considered how much of this structural and taxonomic information is included in transcriptomic data from matched samples. Here, we described fungal communities using both host-derived RNA-Seq and fungal ITS1 DNA amplicon sequencing to compare the outcomes between the methods. We used a panel of root and needle samples from the coniferous tree species Picea abies (Norway spruce) growing in untreated (nutrient-deficient) and nutrient-enriched plots at the Flakaliden forest research site in boreal northern Sweden. We show that the relationship between samples and alpha and beta diversity indicated by the fungal transcriptome is in agreement with that generated by the ITS data, while also identifying a lack of taxonomic overlap due to limitations imposed by current database coverage. Furthermore, we demonstrate how metatranscriptomics data additionally provide biologically informative functional insights. At the community level, there were changes in starch and sucrose metabolism, biosynthesis of amino acids, and pentose and glucuronate interconversions, while processing of organic macromolecules, including aromatic and heterocyclic compounds, was enriched in transcripts assigned to the genus Cortinarius. IMPORTANCE A deeper understanding of microbial communities associated with plants is revealing their importance for plant health and productivity. RNA extracted from plant field samples represents the host and other organisms present. Typically, gene expression studies focus on the plant component or, in a limited number of studies, expression in one or more associated organisms. However, metatranscriptomic data are rarely used for taxonomic profiling, which is currently performed using amplicon approaches. We created an assembly-based, reproducible, and hardware-agnostic workflow to taxonomically and functionally annotate fungal RNA-Seq data obtained from Norway spruce roots, which we compared to matching ITS amplicon sequencing data. While we identified some limitations and caveats, we show that functional, taxonomic, and compositional insights can all be obtained from RNA-Seq data. These findings highlight the potential of metatranscriptomics to advance our understanding of interaction, response, and effect between host plants and their associated microbial communities.

The advent of novel sequencing techniques has unraveled a tremendous diversity on Earth. Genomic data allow us to understand ecology and function of organisms that we would not otherwise know existed. However, major methodological challenges remain, in particular for multicellular organisms with large genomes. Arbuscular mycorrhizal (AM) fungi are important plant symbionts with cryptic and complex multicellular life cycles, thus representing a suitable model system for method development. Here, we report a novel method for large scale, unbiased nuclear sorting, sequencing, and de novo assembling of AM fungal genomes. After comparative analyses of three assembly workflows we discuss how sequence data from single nuclei can best be used for different downstream analyses such as phylogenomics and comparative genomics of single nuclei. Based on analysis of completeness, we conclude that comprehensive de novo genome assemblies can be produced from six to seven nuclei. The method is highly applicable for a broad range of taxa, and will greatly improve our ability to study multicellular eukaryotes with complex life cycles.

Schizophrenia is a common mental disorder with high heritability. It is genetically complex and to date more than a hundred risk loci have been identified. Association of environmental factors and schizophrenia has also been reported, while epigenetic analyses have yielded ambiguous and sometimes conflicting results. Here, we analyzed fresh frozen post-mortem brain tissue from a cohort of 73 subjects diagnosed with schizophrenia and 52 control samples, using the Illumina Infinium HumanMethylation450 Bead Chip, to investigate genome-wide DNA methylation patterns in the two groups. Analysis of differential methylation was performed with the Bioconductor Minfi package and modern machine-learning and visualization techniques, which were shown previously to be successful in detecting and highlighting differentially methylated patterns in case-control studies. In this dataset, however, these methods did not uncover any significant signals discerning the patient group and healthy controls, suggesting that if there are methylation changes associated with schizophrenia, they are heterogeneous and complex with small effect.

In this paper we explore the design space of a self-organizing map (SOM) used for rapid and accurate identification of bacterial genomes. This is an important health care problem because even in Europe, 70% of prescriptions for antibiotics is wrong. The SOM is trained on Next Generation Sequencing (NGS) data and is able to identify the exact strain of bacteria. This is in contrast to conventional methods that require genome assembly to identify the bacterial strain. SOM has been implemented as an synchoros VLSI design and shown to have 3-4 orders better computational efficiency compared to GPUs. To further lower the energy consumption, we exploit the robustness of SOM by successively lowering the resolution to gain further improvements in efficiency and lower the implementation cost without substantially sacrificing the accuracy. We do an in depth analysis of the reduction in resolution vs. loss in accuracy as the basis for designing a system with the lowest cost and acceptable accuracy using NGS data from samples containing multiple bacteria from the labs of one of the co-authors. The objective of this method is to design a bacterial recognition system for battery operated clinical use where the area, power and performance are of critical importance. We demonstrate that with 39% loss in accuracy in 12 hits and 1% in 16 bit representation can yield significant savings in energy and area.

Type 2 diabetes (T2D) mellitus is a complex metabolic disease commonly caused by insulin resistance in several tissues. We performed a matched two-dimensional metabolic screening in tissue samples from 43 multi-organ donors. The intra-individual analysis was assessed across five key metabolic tissues (serum, visceral adipose tissue, liver, pancreatic islets and skeletal muscle), and the inter-individual across three different groups reflecting T2D progression. We identified 92 metabolites differing significantly between non-diabetes and T2D subjects. In diabetes cases, carnitines were significantly higher in liver, while lysophosphatidylcholines were significantly lower in muscle and serum. We tracked the primary tissue of origin for multiple metabolites whose alterations were reflected in serum. An investigation of three major stages spanning from controls, to pre-diabetes and to overt T2D indicated that a subset of lysophosphatidylcholines was significantly lower in the muscle of pre-diabetes subjects. Moreover, glycodeoxycholic acid was significantly higher in liver of pre-diabetes subjects while additional increase in T2D was insignificant. We confirmed many previously reported findings and substantially expanded on them with altered markers for early and overt T2D. Overall, the analysis of this unique dataset can increase the understanding of the metabolic interplay between organs in the development of T2D.

Background: During infection by intracellular pathogens, a highly complex interplay occurs between the infected cell trying to degrade the invader and the pathogen which actively manipulates the host cell to enable survival and proliferation. Many intracellular pathogens pose important threats to human health and major efforts have been undertaken to better understand the host-pathogen interactions that eventually determine the outcome of the infection. Over the last decades, the unicellular eukaryote Dictyostelium discoideum has become an established infection model, serving as a surrogate macrophage that can be infected with a wide range of intracellular pathogens. In this study, we use high-throughput RNA-sequencing to analyze the transcriptional response of D. discoideum when infected with Mycobacterium marinum and Legionella pneumophila. The results were compared to available data from human macrophages.

Results: The majority of the transcriptional regulation triggered by the two pathogens was found to be unique for each bacterial challenge. Hallmark transcriptional signatures were identified for each infection, e.g. induction of endosomal sorting complexes required for transport (ESCRT) and autophagy genes in response to M. marinum and inhibition of genes associated with the translation machinery and energy metabolism in response to L. pneumophila. However, a common response to the pathogenic bacteria was also identified, which was not induced by non-pathogenic food bacteria. Finally, comparison with available data sets of regulation in human monocyte derived macrophages shows that the elicited response in D. discoideum is in many aspects similar to what has been observed in human immune cells in response to Mycobacterium tuberculosis and L. pneumophila.

Conclusions: Our study presents high-throughput characterization of D. discoideum transcriptional response to intracellular pathogens using RNA-seq. We demonstrate that the transcriptional response is in essence distinct to each pathogen and that in many cases, the corresponding regulation is recapitulated in human macrophages after infection by mycobacteria and L. pneumophila. This indicates that host-pathogen interactions are evolutionary conserved, derived from the early interactions between free-living phagocytic cells and bacteria. Taken together, our results strengthen the use of D. discoideum as a general infection model.