Search

Single-cell RNA-sequencing data from multiple time points promises insights into mechanisms controlling differentiation and cell fate decisions at the level of individual cells. Yet, typically only a small number of time points are available. Due to the destructive nature of the sequencing protocol, [for full text, please go to the a.m. URL]

Live imaging techniques, such as two-photon imaging, promise novel insights into cellular activity patterns at a high spatial and temporal resolution. While current deep learning approaches typically focus on specific supervised tasks in the analysis of such data, e.g., learning a segmentation mask as a basis for subsequent signal extraction steps, we investigate how unsupervised generative deep learning can be adapted to obtain interpretable models directly at the level of the video frames. Specifically, we consider variational autoencoders for models that infer a compressed representation of the data in a low-dimensional latent space, allowing for insight into what has been learned. Based on this approach, we illustrate how structural knowledge can be incorporated into the model architecture to improve model fitting and interpretability. Besides standard convolutional neural network components, we propose an architecture for separately encoding the foreground and background of live imaging data. We exemplify the proposed approach with two-photon imaging data from hippocampal CA1 neurons in mice, where we can disentangle the neural activity of interest from the neuropil background signal. Subsequently, we illustrate how to impose smoothness constraints onto the latent space for leveraging knowledge about gradual temporal changes. As a starting point for adaptation to similar live imaging applications, we provide a Jupyter notebook with code for exploration. Taken together, our results illustrate how architecture choices for deep generative models, such as for spatial structure, foreground vs. background, and gradual temporal changes, facilitate a modeling approach that combines the flexibility of deep learning with the benefits of incorporating domain knowledge. Such a strategy is seen to enable interpretable, purely image-based models of activity signals from live imaging, such as for two-photon data.

Background: When considering longitudinal data of individuals, personalised modelling means that each individual receives a customised dynamic model. Yet, biomedical time-series data from individuals are often characterised by a sparse, highly irregular time grid of measurements and individual-specific[for full text, please go to the a.m. URL], 65th Annual Meeting of the German Association for Medical Informatics, Biometry and Epidemiology (GMDS), Meeting of the Central European Network (CEN: German Region, Austro-Swiss Region and Polish Region) of the International Biometric Society (IBS)

Resumen del póster presentado en Granada, entre el 14 y el 16 de noviembre de 2018., Fruit quality and crop productivity in legume species are critically influenced by maturation and dehiscence developmental processes (Tang et al., 2013). These stages of fruit development are well characterized in the Arabidopsis plant model but not in legumes, as is the case of common bean Phaseolus vulgaris (Li and Olsen, 2016). Here, we used RNA¿seq to compare immature and mature stages of pod development in two accessions (PMB0225 and PHA1037) of this species, which also differ in other important agronomical traits such as dehiscence and fiber content. The most recent update of the Phaseolus vulgaris reference genome was used for the alignment of RNA¿seq short reads, then computing fold changes among the different developmental stages. In this way, we were able to derive a list of differentially expressed genes (DEG) that are associated with pod maturation in both accessions, mainly involved in transmembrane transporter activity, carbohydrate metabolism and photosynthesis. Furthermore, among DEGs between dehiscent and indehiscent fruits we found genes related to oxidation¿reduction and lipid metabolic processes., This work was financially supported by the Ministerio de Economía y Competitividad (AGL 2014-51809-R, AGL2015-64991-C3-1-R, AGL2017-88702-C2-2-R, AGL2017-88174-R) and UE-FEDER Program.

Resumen del póster presentado en la "8th International Conference on Legume Genetics and Genomics (ICLGG)", que tuvo lugar en Siófok, Hungría, entre el 18 y el 22 de septiembre de 2017., Fruit maturation and dehiscence are important developmental processes in legume species as they influence fruit quality and crop productivity (Tang et al., 2013). Despite that key regulators of fruit development have been characterized in the model species Arabidopsis thaliana, the genetic basis of fruit maturation and dehiscence remains unknown in agricultural species like common bean (Li and Olsen, 2016). Recent advances in genetic, molecular and physiological research, mostly brought about by the implementation of high-throughput and next generation sequencing (NGS) technologies have begun to decipher the genetic networks regulating fruit development. With the aim to unravel the transcriptomic changes associated to pod maturation and dehiscence, we used a powerful RNA-seq bioinformatics protocol to compared immature and mature pods of two accessions, which also differed in deshiscence and fiber content. RNA-seq short reads were aligned to the most recent update of the Phaseolus vulgaris reference genome, then computing the fold changes in assembled transcript expression among different developmental stages. Using Phytozome gene annotation for common bean, a significant number of differentially expressed genes (DEGs) were associated with pod maturation in both accessions, which were mainly involved in transmembrane transporter activity, carbohydrate metabolism and photosynthesis. Furthermore, among DEGs between dehiscent and indehiscent fruits highlights genes related to oxidation-reduction and lipid metabolic processes. This work provides comprehensive insights into pod maturation and dehiscence, identifying gene expression profiles with major roles on these developmental processes., This work was financially supported by the Ministerio de Economía y Competitividad (AGL2014-51809-R and AGL2015-64991-C3-1-R) and UE-FEDER Program.

Nasopharyngeal carcinoma (NPC) is an aggressive tumour associated with Epstein-Barr virus (EBV), which has a poor prognosis when detected late. Easy methods for detecting NPC at earlier stages are highly needed. Host and viral miRNAs are perturbed in NPC, a fraction of which are included in extracellular vesicles and/or in ribonucleoprotein complexes that enter into the circulation. Measuring circulating miRNAs may provide a robust and direct insight of the oncogenic process. Here we analysed whether host and viral circulating miRNA fingerprints may reflect tumour burden and behaviour. Defined circulating microRNAs were analysed from total serum and plasma (NPC, non-NPC patients and healthy controls). Selected NPC sera and plasma were fractionated by size-exclusion chromatography into extracellular vesicle and lipoprotein fractions. Exosome-like vesicles released from EBV-infected NPC were characterized by western blot and transmission electron microscopy. Deep RNA sequencing (RNAseq) was done for NPC cell lines, exosomes and selected sera. EBV DNA load was measured as well. Viral miRNA (mirBART7, 9, 13), host miRNA (miR16, 21, 155) and small RNAs (EBER1 and vaultRNA) were detected by stemloop reverse transcription-polymerase chain reaction (RT-PCR). We detected viral miRNAs in NPC sera and plasma which levels seem to increase with TNM stage despite absence of detectable EBV DNA in some. EBER1 RNA was absent in virtually all NPC sera and plasma, despite high levels in NPC tissues. Cellular miR155, which expression is induced by EBV-LMP1, was significantly elevated in NPC when compared to controls. Fractionation from NPC sera and plasma revealed that miR155 and vault RNA are selectively enriched in extracellular vesicles, whereas viral miRNAs are more randomly distributed. The pattern of tetraspanin (CD63, CD81) protein expression is heterogeneous in NPC patients compared to healthy controls. Further, RNAseq and RT-PCR analysis are being performed to determine the diagnostic potential of circulating miRNAs for NPC.

In this work, we present a model describing the boron redistribution during laser thermal annealing in the melting regime based on the adsorption of boron atoms at the solid-liquid interface. To validate the model, we performed SIMS measurements on silicon samples implanted with boron with an energy of 3 keV and doses of 3×1013 cm-2 and 4×1014 cm-2 annealed with a XeCl excimer laser with a wavelength of 308 nm, a pulse duration of 160 ns, and up to 10 consecutive pulses. After calibration, our model is able to reproduce the measured profiles for the different process conditions.

BF3 plasma immersion ion implantation (PIII) is a promising technique in the race for highly boron doped P+/N ultra-shallow junctions (USJs) in complementary metal oxide semiconductor (CMOS) silicon technologies. Implantation conditions used in BF3 PIII lead to high super-saturations (1 × 1020 cm-3) of both boron and silicon interstitial atoms in the implantation region. In such conditions, very large loop-shaped boroninterstitial clusters (BICs) are formed during subsequent thermal anneals, as confirmed by transmission electron microscopy (TEM) measurements. In this study, amorphizing BF3 PIII implants (10 keV, 5 × 1015 cm-2) followed by different thermal anneals were carried out in order to investigate the large BICs precipitation. A "large BICs" model based on moments approach allowed to reproduce the experimental data including boron diffusion profiles (obtained by SIMS) and boron electrical activation obtained by Hall effect sheet resistance measurements.