Department of Electrical and Computer Engineering

We previously showed that I Cl,swell is activated by ROS generation via NADPH oxidase (NOX) and the mitochondrial electron transport chain (ETC). Sphingolipid signaling is implicated in channel regulation. Here we examined the role of ceramide in the modulation of I Cl,swell . Under isosmotic conditions, the addition of exogenous C 2 -ceramide (C 2 -Cer, 2 fM) increased Cl À current density by 0.7 5 0.1 pA/pF at þ60 mV after 10 min (n ¼ 11, P<0.01). DCPIB (10 fM), a highly selective I Cl,swell antagonist, inhibited C 2 -Cer-induced I Cl,swell by 76 5 8% (n ¼

enhancing the mitochondrial ROS production with either antimycin or rotenone. Oscillations of IKATP could be either initiated or potentiated upon rapid, but not slow, transition to near-anoxia and they were closely paralleled by depolarization of delta Psi, indicative of a transient inability of the F1F0-ATPase to keep delta Psi. At elevated oxidative stress, rapid transition to near-anoxia caused a burst of H2DCF oxidation which correlated with an increased rate of IKATP activation. These results show that metabolic oscillations occur in cardiomyocytes at nearanoxia and that these oscillations are controlled by mitochondria through the rate of ATP hydrolysis which in turn depends on ROS production.

IsoþBayK-induced EADs by fast pacing rates and Ca accumulation; 2) The upstrokes of H 2 O 2 -induced EADs preceded the initiations of Ca aftertransients (CaATs), while the upstrokes of IsoþBayK-induced EADs occurred after the initiations of CaATs; 3) The EAD take-off potentials were within a narrow range between þ20 and À20 mV in H 2 O 2 model, while they were randomly distributed in a wide range between þ30 and À60 mV, corresponding to concomitant Ca waves in IsoþBayK model; 4) When the cell membrane clamp was switched from current clamp to voltage clamp using a normal AP waveform without EAD, CaATs/Ca waves disappeared in H 2 O 2 model, but persisted in IsoþBayK model. 5) The I Ca,L blocker nifedipine (10 mM), eliminated EADs and CaATs in both models. 6) SEA400 (2 mM), a selective blocker of Na-Ca exchange current, suppressed EADs in both models and H 2 O 2 -inducd CaAT, but exerted less effect on Ca wave in IsoþBayK model. We conclude that the mechanisms for EADs vary depending on their causative factors. While reactivation of I Ca,L plays a predominant role in EAD genesis by H 2 O 2 , spontaneous Ca waves are predominant cause under Ca-overload conditions (e.g. in IsoþBayK model). Rationale: In sinoatrial node cells (SANC), Ca 2þ activates adenylate cyclase (AC) to generate a high basal level of cAMP-mediated/protein kinase A (PKA)-dependent phosphorylation of Ca 2þ cycling proteins. This results in spontaneous sarcoplasmic reticulum-generated rhythmic Ca 2þ oscillations during diastolic depolarization, that not only ignite the surface membrane to generate rhythmic action potentials (APs), but, in a feed-forward manner, also activate AC/PKA signaling. ATP is consumed to produce cAMP, to pump Ca 2þ and to contract. Objective: Since nature efficiently links ATP-demand to ATP production, we hypothesized that (1) both basal ATP supply and demand in SANC would be Ca 2þ -cAMP/PKA dependent; and (2) due to its feed-forward nature, a decrease in flux through the Ca 2þ -cAMP/PKA signaling axis will reduce the basal ATP level. Methods and Results: Graded reduction of basal Ca 2þ -cAMP/PKA signaling in rabbit SANC, produced graded ATP depletion (r 2 =0.96), and reduced O 2 consumption and flavoprotein fluorescence. Neither inhibition of glycolysis, nor selectively blocking contraction reduced the ATP level. Specific inhibition of mitochondrial Ca 2þ flux was without effect, indicating that the cAMP/PKA component rather than Ca 2þ directly, links the Ca 2þ /cAMP-PKA signaling to ATP production. Conclusions: Feed-forward basal Ca 2þ -cAMP/PKA signaling both consumes ATP to drive spontaneous APs in SANC and is tightly linked to mitochondrial ATP production. Interfering with Ca 2þ -cAMP/PKA signaling to reduce the SANC ATP demand also ''pulls the plug'' on SANC ATP supply. This distinctly differs from ventricular myocytes, which lack this feed-forward basal cAMP/PKA signaling, and in which ATP level remains constant when the demand changes.

Fluorescence imaging of transmembrane voltage-sensitive dyes is used to study electrical activation in cardiac tissue. However, the fluorescence signals, typically, have low SNRs and may be contaminated with motion artifact. In this report, we introduce a new processing approach for fluoresced transmembrane potentials (fTmps) that is based upon a discrete wavelet transform. We show how fTmp signals can be decomposed and reconstructed to form three subsignals that contain signal noise (noise signal), the early depolarization phase of the action potential (rTmp signal), and motion artifact (rMA signal). A coiflet4 wavelet is used for fTmp decomposition and reconstruction of these subsignals. Results using fTmp signals that are contaminated with motion artifact indicate that the approach is a useful processing step to remove baseline drift, reduce noise, and reveal wavefronts. It streamlines the preprocessing of fTmps for the subsequent measurement of activation times and conduction velocities. It is a promising approach for studying wavefronts without aggressive mechanical tissue constraint or electromechanical uncoupling agents and is, useful for single-camera systems that do not provide for ratiometric imaging.

"Grid computing has recently emerged as a response to the growing demand for computing resources exhibited by the scientific and researcher communities.
The Grid has witnessed rapid growth in the past few years. Several countries have adopted this technology as a basic computing infrastructure for scientific and research disciplines.
We present a report on Grid activities in Morocco in the last three years as a consequence of an established collaboration between our three institutions.
A national grid initiative, called MaGrid, has been setup which aims at deploying a robust, flexible and sustainable Grid infrastructure.
This initiative proved to be a catalyst for other Moroccan universities and research centers in the country with a growing interest to collaborate in Grid activities. Moreover the involvement of Morocco in the EumedGrid project, in the Mediterranean region, will open new collaboration efforts
in terms of the Grid."

"Ce papier expose brièvement, la technologie de la Grille de Calcul et son apport essentiel à toute la communauté
scientifique. La collaboration Belgo-Maorcaine à travers des associations de biologistes, de physiciens et de chercheurs
en nouvelles technologies, sera abordée par la suite, et montrera le rôle bénéfique de ces structures dans le transfert
technologique entre les deux pays. Ensuite, la collaboration réussie, dans le domaine de l’investigation en Grille de
Calcul, entre la Belgique et le Maroc sera abordée. Elle est, en effet, le résultat d’un travail collectif et assidu, mené
conjointement, depuis 2003, par l’Université Libre de Bruxelles (ULB), l’Université Abdelmalek Essaadi (UAE) et le
Centre National pour la Recherche Scientifique et Technique (CNRST)."

"The Weather Research and Forecast (WRF) is a specific program for forecasting and research. WRF processing is composed of three steps: preprocessing, core model execution and post-processing.
WRF/Chem is the WRF coupled with Chemistry. The model simulates the coupling between dynamics, radiation and chemistry, and is used for investigation of regional-scale air quality, field program analysis, and cloud-scale interactions between clouds and chemistry. WRF/Chem can predict transport of atmospheric constituents and radiation of O3 and UV.
The WRF processing requires high execution time and storage space. The porting to Grid Computing can be very advantageous since the model supports hybrid distributed and shared memory computations.
The Grid platforms used for this study are MaGrid and EumedGrid. MaGrid is linked to EumedGrid, and users can have access to appropriate CPU and storage resources. The Grid Computing abundant resources enable multiple scenarios to be run in parallel, and the MPI parallel programming model allows execution time to decrease substantially with increasing the number of Worker Nodes. In addition, for most simulations, numerical grid can be more refined especially in sensitive regions, leading to a higher degree of results accuracy.
In this work, the WRF is used to forecast the weather, and to predict the effects of ships crossing the Gibraltar Strait on Ozone levels in the atmosphere.
WRF model has been successfully implemented in MaGrid/EumedGrid, for operational and development purposes. The WPS execution was sequential while WRF was parallel on distributed memory according to MPI. The overall speedup increases substantially when CPU number increases."

Air pollution is one of the most serious problems the world faces today. It is highly necessary to monitor pollutants in real-time to anticipate and reduce damages caused in several fields of activities. Likewise, it is necessary to provide decision makers with useful and updated environmental data. As a solution to a part of the above-mentioned necessities, we developed a Java-based application software to collect, process and visualize several environmental and pollution data, acquired from the Mediterranean Dialog earth Observatory (MDEO) platform [1]. This application will amass data of Morocco area from EUMETSAT satellites, and will decompress, filter and classify the received datasets. Then we will use the processed data to build an interactive environmental real-time map of Morocco. This should help finding out potential correlations between pollutants and emitting sources.

We propose a deep learning method for Atmospheric Ozone Interpolation. Our method directly learns an end-to-end mapping between classically interpolated satellite ozone images and the real ozone measurements. The model architecture represents a deep stack of convolutions (CNN) that takes the already interpolated images (Using the classical state-of-theart interpolation method) as Input and outputs a more precise Interpolation of the Region of Interest. Our deep CNN has a lightweight structure, yet demonstrates state-of-the-art interpolation quality, and achieves optimal data processing latency (∆T) for production-ready near-real-time Atmospheric Image Interpolation, which has a big advantage over the state of the art classical interpolation algorithms. We explore different network structures and parameter settings to achieve trade-offs between performance and speed. This method showcases the potential applications of deep learning in Remote Sensing and Climate Science. 1 Introduction Remote sensing data represents a big opportunity in the field of climate science and atmospheric environmental research Ustin (2004), satellites with high performing sensors provide rich and continuous streams of data about weather, agriculture, pollution, oceans among many other natural phenomena Joshi et al. (2016). This paper focuses on Atmospheric Ozone Monitoring using remote sensing data, EUMETSAT is responsible for streaming rich data-sets in short intervals of time to the many users they have (among them us at MDEO El Amrani et al. (2012)), EUMETSAT offers ozone scans over Africa and the whole 0-degree face of the earth, a lot of this data has missing or damaged pixel ozone measurement values due to the many natural phenomena that prevent MSG's SEVIRI sensor from capturing valid ozone measurements. On the other side, 2-D Interpolation is a classical problem in Mathematics and Computer Vision, and state-of-the-art 2-D Interpolation algorithms are well established and researched. However, research in Satellite Image Interpolation is not common, and trade-offs between Interpolation speed and Accuracy for different types of remote sensing image data might be hard to document. This paper studies the trade-offs as well as the performance of Convolution-based deep neural models and benchmarks a deep CNN LeCun et al. (1990) against the classical state of the art 2-D interpolation algorithms.

In this paper, we describe the Cloud Computing basic compute resources scheduling and allocation algorithms, in addition to the working mechanism. This paper also presents a number of experiments conducted based on CloudSim simulation toolkit in order to assess and evaluate the performance of these scheduling algorithms on Cloud Computing like infrastructure. Furthermore, we introduced and explained the CloudSim simulator design, architecture and proposed two new scheduling algorithms to enhance the existent ones and highlight the weaknesses and/or effectiveness of these algorithms.