An accurate exchange rate forecasting and its decision-making to buy or sell are critical issues in the Forex market. Short-term currency rate forecasting is a challenging task due to its inherent characteristics, which include high volatility, trend, noise, and market shocks. We propose a novel deep learning architecture consisting of an adaptive activation function selection mechanism to achieve higher predictive accuracy. The proposed architecture is composed of seven neural networks that have different activation functions as well as softmax layer and multiplication layer with a skip connection, which are used to generate the dynamic importance weights that decide which activation function is preferred. In addition, we introduce an extended Min-Max smoothing technique to further normalize financial time series that have non-stationary properties. In our experimental evaluation, the results showed that our proposed model not only outperforms deep neural network baselines but also other classic machine learning approaches. The extended Min-Max smoothing technique is step towards forecasting non-stationary financial time series with deep neural networks.

Phycobilisome (PBS) is the main light-harvesting antenna in cyanobacteria and red algae. How PBS transfers the light energy to photosystem II (PSII) remains to be elucidated. Here we report the in situ structure of the PBS–PSII supercomplex from Porphyridium purpureum UTEX 2757 using cryo-electron tomography and subtomogram averaging. Our work reveals the organized network of hemiellipsoidal PBS with PSII on the thylakoid membrane in the native cellular environment. In the PBS–PSII supercomplex, each PBS interacts with six PSII monomers, of which four directly bind to the PBS, and two bind indirectly. Additional three ‘connector’ proteins also contribute to the connections between PBS and PSIIs. Two PsbO subunits from adjacent PSII dimers bind with each other, which may promote stabilization of the PBS–PSII supercomplex. By analyzing the interaction interface between PBS and PSII, we reveal that α^LCM and ApcD connect with CP43 of PSII monomer and that α^LCM also interacts with CP47' of the neighboring PSII monomer, suggesting the multiple light energy delivery pathways. The in situ structures illustrate the coupling pattern of PBS and PSII and the arrangement of the PBS–PSII supercomplex on the thylakoid, providing the near-native 3D structural information of the various energy transfer from PBS to PSII.

Combination drug therapy is an efficient way to treat complicated diseases. Drug-drug interaction (DDI) is an important research topic in this therapy as patient safety is a problem when two or more drugs are taken at the same time. Traditionally, in vitro experiments and clinical trials are common ways to determine DDIs. However, these methods cannot meet the requirements of large-scale tests. It is an alternative way to develop computational methods for predicting DDIs. Although several previous methods have been proposed, they always need several types of drug information, limiting their applications. In this study, we proposed a simple computational method to predict DDIs. In this method, drugs were represented by their fingerprint features, which are most widely used in investigating drug-related problems. These features were refined by three models, including addition, subtraction, and Hadamard models, to generate the representation of DDIs. The powerful classification algorithm, random forest, was picked up to build the classifier. The results of two types of tenfold cross-validation on the classifier indicated good performance for discovering novel DDIs among known drugs and acceptable performance for identifying DDIs between known drugs and unknown drugs or among unknown drugs. Although the classifier adopted a sample scheme to represent DDIs, it was still superior to other methods, which adopted features generated by some advanced computer algorithms. Furthermore, a user-friendly web-server, named DDIPF (http://106.14.164.77:5004/DDIPF/), was developed to implement the classifier.

Selective macroautophagy (hereafter referred to as autophagy) describes a process in which cytosolic material is engulfed in a double membrane organelle called an autophagosome. Autophagosomes are carriers responsible for delivering their content to a lytic compartment for destruction. The cargo can be of diverse origin, ranging from macromolecular complexes to protein aggregates, organelles, and even invading pathogens. Each cargo is unique in composition and size, presenting different challenges to autophagosome biogenesis. Among the largest cargoes targeted by the autophagy machinery are intracellular bacteria, which can, in the case of Salmonella, range from 2 to 5 μm in length and 0.5 to 1.5 μm in width. How phagophores form and expand on such a large cargo remains mechanistically unclear. Here, we used HeLa cells infected with an auxotrophic Salmonella to study the process of phagophore biogenesis using in situ correlative cryo-ET. We show that host cells generate multiple phagophores at the site of damaged Salmonella -containing vacuoles (SCVs). The observed double membrane structures range from disk-shaped to expanded cup-shaped phagophores, which have a thin intermembrane lumen with a dilating rim region and expand using the SCV, the outer membrane of Salmonella , or existing phagophores as templates. Phagophore rims establish different forms of contact with the endoplasmic reticulum (ER) via structurally distinct molecular entities for membrane formation and expansion. Early omegasomes correlated with the marker Double-FYVE domain-Containing Protein 1 (DFCP1) are observed in close association with the ER without apparent membrane continuity. Our study provides insights into the formation of phagophores around one of the largest selective cargoes.