Electrocardiograms (ECG) record the heart activity and are the most common and reliable method to detect cardiac arrhythmias, such as atrial fibrillation (AFib). Lately, many commercially available devices such as smartwatches are offering ECG monitoring. Therefore, there is increasing demand for designing deep learning models with the perspective to be physically implemented on these small portable devices with limited energy supply. In this paper, a workflow for the design of small, energy-efficient recurrent convolutional neural network (RCNN) architecture for AFib detection is proposed. However, the approach can be well generalized to every type of long time series. In contrast to previous studies, that demand thousands of additional network neurons and millions of extra model parameters, the logical steps for the generation of a CNN with only 114 trainable parameters are described. The model consists of a small segmented CNN in combination with an optimal energy classifier. The architectural decisions are made by using the energy consumption as a metric in an equally important way as the accuracy. The optimisation steps are focused on the software which can be embedded afterwards on a physical chip. Finally, a comparison with some previous relevant studies suggests that the widely used huge CNNs for similar tasks are mostly redundant and unessentially computationally expensive.
KDEL receptors (KDELRs) represent transmembrane proteins of the secretory pathway which regulate the retention of soluble ER-residents as well as retrograde and anterograde vesicle trafficking. In addition, KDELRs are involved in the regulation of cellular stress response and ECM degradation. For a deeper insight into KDELR1 specific functions, we characterised a KDELR1-KO cell line (HAP1) through whole transcriptome analysis by comparing KDELR1-KO cells with its respective HAP1 wild-type. Our data indicate more than 300 significantly and differentially expressed genes whose gene products are mainly involved in developmental processes such as cell adhesion and ECM composition, pointing out to severe cellular disorders due to a loss of KDELR1. Impaired adhesion capacity of KDELR1-KO cells was further demonstrated through in vitro adhesion assays, while collagen- and/or laminin-coating nearly doubled the adhesion property of KDELR1-KO cells compared to wild-type, confirming a transcriptional adaptation to improve or restore the cellular adhesion capability. Perturbations within the secretory pathway were verified by an increased secretion of ER-resident PDI and decreased cell viability under ER stress conditions, suggesting KDELR1-KO cells to be severely impaired in maintaining cellular homeostasis.
Background: Bidirectional promoters (BPs) are prevalent in eukaryotic genomes. However, it is poorly understood how the cell integrates different epigenomic information, such as transcription factor (TF) binding and chromatin marks, to drive gene expression at BPs. Single-cell sequencing technologies are revolutionizing the field of genome biology. Therefore, this study focuses on the integration of single-cell RNA-seq data with bulk ChIP-seq and other epigenetics data, for which single-cell technologies are not yet established, in the context of BPs.
Results: We performed integrative analyses of novel human single-cell RNA-seq (scRNA-seq) data with bulk ChIP-seq and other epigenetics data. scRNA-seq data revealed distinct transcription states of BPs that were previously not recognized. We find associations between these transcription states to distinct patterns in structural gene features, DNA accessibility, histone modification, DNA methylation and TF binding profiles.
Conclusions: Our results suggest that a complex interplay of all of these elements is required to achieve BP-specific transcriptional output in this specialized promoter configuration. Further, our study implies that novel statistical methods can be developed to deconvolute masked subpopulations of cells measured with different bulk epigenomic assays using scRNA-seq data.
An ontology-based method for assessing batch effect adjustment approaches in heterogeneous datasets
(2018)
Motivation: International consortia such as the Genotype-Tissue Expression (GTEx) project, The Cancer Genome Atlas (TCGA) or the International Human Epigenetics Consortium (IHEC) have produced a wealth of genomic datasets with the goal of advancing our understanding of cell differentiation and disease mechanisms. However, utilizing all of these data effectively through integrative analysis is hampered by batch effects, large cell type heterogeneity and low replicate numbers. To study if batch effects across datasets can be observed and adjusted for, we analyze RNA-seq data of 215 samples from ENCODE, Roadmap, BLUEPRINT and DEEP as well as 1336 samples from GTEx and TCGA. While batch effects are a considerable issue, it is non-trivial to determine if batch adjustment leads to an improvement in data quality, especially in cases of low replicate numbers.
Results: We present a novel method for assessing the performance of batch effect adjustment methods on heterogeneous data. Our method borrows information from the Cell Ontology to establish if batch adjustment leads to a better agreement between observed pairwise similarity and similarity of cell types inferred from the ontology. A comparison of state-of-the art batch effect adjustment methods suggests that batch effects in heterogeneous datasets with low replicate numbers cannot be adequately adjusted. Better methods need to be developed, which can be assessed objectively in the framework presented here.
Cancer-associated fibroblasts (CAFs) in the tumor microenvironment contribute to all stages of tumorigenesis and are usually considered to be tumor-promoting cells. CAFs show a remarkable degree of heterogeneity, which is attributed to developmental origin or to local environmental niches, resulting in distinct CAF subsets within individual tumors. While CAF heterogeneity is frequently investigated in late-stage tumors, data on longitudinal CAF development in tumors are lacking. To this end, we used the transgenic polyoma middle T oncogene-induced mouse mammary carcinoma model and performed whole transcriptome analysis in FACS-sorted fibroblasts from early- and late-stage tumors. We observed a shift in fibroblast populations over time towards a subset previously shown to negatively correlate with patient survival, which was confirmed by multispectral immunofluorescence analysis. Moreover, we identified a transcriptomic signature distinguishing CAFs from early- and late-stage tumors. Importantly, the signature of early-stage CAFs correlated well with tumor stage and survival in human mammary carcinoma patients. A random forest analysis suggested predictive value of the complete set of differentially expressed genes between early- and late-stage CAFs on bulk tumor patient samples, supporting the clinical relevance of our findings. In conclusion, our data show transcriptome alterations in CAFs during tumorigenesis in the mammary gland, which suggest that CAFs are educated by the tumor over time to promote tumor development. Moreover, we show that murine CAF gene signatures can harbor predictive value for human cancer.
Several studies suggested that transcription factor (TF) binding to DNA may be impaired or enhanced by DNA methylation. We present MeDeMo, a toolbox for TF motif analysis that combines information about DNA methylation with models capturing intra-motif dependencies. In a large-scale study using ChIP-seq data for 335 TFs, we identify novel TFs that are affected by DNA methylation. Overall, we find that CpG methylation decreases the likelihood of binding for the majority of TFs. For a considerable subset of TFs, we show that intra-motif dependencies are pivotal for accurately modelling the impact of DNA methylation on TF binding.
Long non-coding RNAs (lncRNAs) can act as regulatory RNAs which, by altering the expression of target genes, impact on the cellular phenotype and cardiovascular disease development. Endothelial lncRNAs and their vascular functions are largely undefined. Deep RNA-Seq and FANTOM5 CAGE analysis revealed the lncRNA LINC00607 to be highly enriched in human endothelial cells. LINC00607 was induced in response to hypoxia, arteriosclerosis regression in non-human primates and also in response to propranolol used to induce regression of human arteriovenous malformations. siRNA knockdown or CRISPR/Cas9 knockout of LINC00607 attenuated VEGF-A-induced angiogenic sprouting. LINC00607 knockout in endothelial cells also integrated less into newly formed vascular networks in an in vivo assay in SCID mice. Overexpression of LINC00607 in CRISPR knockout cells restored normal endothelial function. RNA- and ATAC-Seq after LINC00607 knockout revealed changes in the transcription of endothelial gene sets linked to the endothelial phenotype and in chromatin accessibility around ERG-binding sites. Mechanistically, LINC00607 interacted with the SWI/SNF chromatin remodeling protein BRG1. CRISPR/Cas9-mediated knockout of BRG1 in HUVEC followed by CUT&RUN revealed that BRG1 is required to secure a stable chromatin state, mainly on ERG-binding sites. In conclusion, LINC00607 is an endothelial-enriched lncRNA that maintains ERG target gene transcription by interacting with the chromatin remodeler BRG1.
Understanding the role of short-interfering RNA (siRNA) in diverse biological processes is of current interest and often approached through small RNA sequencing. However, analysis of these datasets is difficult due to the complexity of biological RNA processing pathways, which differ between species. Several properties like strand specificity, length distribution, and distribution of soft-clipped bases are few parameters known to guide researchers in understanding the role of siRNAs. We present RAPID, a generic eukaryotic siRNA analysis pipeline, which captures information inherent in the datasets and automatically produces numerous visualizations as user-friendly HTML reports, covering multiple categories required for siRNA analysis. RAPID also facilitates an automated comparison of multiple datasets, with one of the normalization techniques dedicated for siRNA knockdown analysis, and integrates differential expression analysis using DESeq2.
Electrocardiograms (ECG) record the heart activity and are the most common and reliable method to detect cardiac arrhythmias, such as atrial fibrillation (AFib). Lately, many commercially available devices such as smartwatches are offering ECG monitoring. Therefore, there is increasing demand for designing deep learning models with the perspective to be physically implemented on these small portable devices with limited energy supply. In this paper, a workflow for the design of small, energy-efficient recurrent convolutional neural network (RCNN) architecture for AFib detection is proposed. However, the approach can be well generalized to every type of long time series. In contrast to previous studies, that demand thousands of additional network neurons and millions of extra model parameters, the logical steps for the generation of a CNN with only 114 trainable parameters are described. The model consists of a small segmented CNN in combination with an optimal energy classifier. The architectural decisions are made by using the energy consumption as a metric in an equally important way as the accuracy. The optimization steps are focused on the software which can be embedded afterwards on a physical chip. Finally, a comparison with some previous relevant studies suggests that the widely used huge CNNs for similar tasks are mostly redundant and unessentially computationally expensive.
For medicine to fulfill its promise of personalized treatments based on a better understanding of disease biology, computational and statistical tools must exist to analyze the increasing amount of patient data that becomes available. A particular challenge is that several types of data are being measured to cope with the complexity of the underlying systems, enhance predictive modeling and enrich molecular understanding.
Here we review a number of recent approaches that specialize in the analysis of multimodal data in the context of predictive biomedicine. We focus on methods that combine different OMIC measurements with image or genome variation data. Our overview shows the diversity of methods that address analysis challenges and reveals new avenues for novel developments.