Deep Siamese Metric Learning: A Highly Scalable Approach to Searching Unordered Sets of Trajectories

This work proposes metric learning for fast similarity-based scene retrieval of unstructured ensembles of trajectory data from large databases. We present a novel representation learning approach using Siamese Metric Learning that approximates a distance preserving low-dimensional representation and that learns to estimate reasonable solutions to the assignment problem. To this end, we employ a Temporal Convolutional Network architecture that we extend with a gating mechanism to enable learning from sparse data, leading to solutions to the assignment problem exhibiting varying degrees of sparsity. Our experimental results on professional soccer tracking data provides insights on learned features and embeddings, as well as on generalization, sensitivity, and network architectural considerations. Our low approximation errors for learned representations and the interactive performance with retrieval times several magnitudes smaller shows that we outperform previous state of the art.

Make More Connections: Urban Traffic Flow Forecasting with Spatiotemporal Adaptive Gated Graph Convolution Network

Urban traffic flow forecasting is a critical issue in intelligent transportation systems. Due to the complexity and uncertainty of urban road conditions, how to capture the dynamic spatiotemporal correlation and make accurate predictions is very challenging. In most of existing works, urban road network is often modeled as a fixed graph based on local proximity. However, such modeling is not sufficient to describe the dynamics of the road network and capture the global contextual information. In this paper, we consider constructing the road network as a dynamic weighted graph through attention mechanism. Furthermore, we propose to seek both spatial neighbors and semantic neighbors to make more connections between road nodes. We propose a novel Spatiotemporal Adaptive Gated Graph Convolution Network ( STAG-GCN ) to predict traffic conditions for several time steps ahead. STAG-GCN mainly consists of two major components: (1) multivariate self-attention Temporal Convolution Network ( TCN ) is utilized to capture local and long-range temporal dependencies across recent, daily-periodic and weekly-periodic observations; (2) mix-hop AG-GCN extracts selective spatial and semantic dependencies within multi-layer stacking through adaptive graph gating mechanism and mix-hop propagation mechanism. The output of different components are weighted fused to generate the final prediction results. Extensive experiments on two real-world large scale urban traffic dataset have verified the effectiveness, and the multi-step forecasting performance of our proposed models outperforms the state-of-the-art baselines.

Rapid Bursts of Magnetically Gated Accretion in the Intermediate Polar V1025 Cen

Magnetically gated accretion has emerged as a proposed mechanism for producing extremely short, repetitive bursts of accretion onto magnetized white dwarfs in intermediate polars (IPs), but this phenomenon has not been detected previously in a confirmed IP. We report the 27 day TESS light curve of V1025 Cen, an IP that shows a remarkable series of 12 bursts of accretion, each lasting for less than 6 hours. The extreme brevity of the bursts and their short recurrence times (∼1–3 days) are incompatible with the dwarf-nova instability, but they are natural consequences of the magnetic gating mechanism developed by Spruit and Taam to explain the Type II bursts of the accreting neutron star known as the Rapid Burster. In this model, the accretion flow piles up at the magnetospheric boundary and presses inward until it couples with the star’s magnetic field, producing an abrupt burst of accretion. After each burst, the reservoir of matter at the edge of the magnetosphere is replenished, leading to cyclical bursts of accretion. A pair of recent studies applied this instability to the suspected IPs MV Lyr and TW Pic, but the magnetic nature of these two systems has not been independently confirmed. In contrast, previous studies have unambiguously established the white dwarf in V1025 Cen to be significantly magnetized. The detection of magnetically gated bursts in a confirmed IP therefore validates the extension of the Spruit and Taam instability to magnetized white dwarfs.

Integrative Study of the Structural and Dynamical Properties of a KirBac3.1 Mutant: Functional Implication of a Highly Conserved Tryptophan in the Transmembrane Domain

ATP-sensitive potassium (K-ATP) channels are ubiquitously expressed on the plasma membrane of cells in several organs, including the heart, pancreas, and brain, and they govern a wide range of physiological processes. In pancreatic β-cells, K-ATP channels composed of Kir6.2 and SUR1 play a key role in coupling blood glucose and insulin secretion. A tryptophan residue located at the cytosolic end of the transmembrane helix is highly conserved in eukaryote and prokaryote Kir channels. Any mutation on this amino acid causes a gain of function and neonatal diabetes mellitus. In this study, we have investigated the effect of mutation on this highly conserved residue on a KirBac channel (prokaryotic homolog of mammalian Kir6.2). We provide the crystal structure of the mutant KirBac3.1 W46R (equivalent to W68R in Kir6.2) and its conformational flexibility properties using HDX-MS. In addition, the detailed dynamical view of the mutant during the gating was investigated using the in silico method. Finally, functional assays have been performed. A comparison of important structural determinants for the gating mechanism between the wild type KirBac and the mutant W46R suggests interesting structural and dynamical clues and a mechanism of action of the mutation that leads to the gain of function.

Proton transfer pathway in anion channelrhodopsin-1

Anion channelrhodopsin from Guillardia theta (GtACR1) has Asp234 (3.2 Å) and Glu68 (5.3 Å) near the protonated Schiff base. Here, we investigate mutant GtACR1s (e.g., E68Q/D234N) expressed in HEK293 cells. The influence of the acidic residues on the absorption wavelengths was also analyzed using a quantum mechanical/molecular mechanical approach. The calculated protonation pattern indicates that Asp234 is deprotonated and Glu68 is protonated in the original crystal structures. The D234E mutation and the E68Q/D234N mutation shorten and lengthen the measured and calculated absorption wavelengths, respectively, which suggests that Asp234 is deprotonated in the wild-type GtACR1. Molecular dynamics simulations show that upon mutation of deprotonated Asp234 to asparagine, deprotonated Glu68 reorients toward the Schiff base and the calculated absorption wavelength remains unchanged. The formation of the proton transfer pathway via Asp234 toward Glu68 and the disconnection of the anion conducting channel are likely a basis of the gating mechanism.

Structure of a TRAPPII-Rab11 activation intermediate reveals GTPase substrate selection mechanisms

Rab1 and Rab11 are essential regulators of the eukaryotic secretory and endocytic recycling pathways. The TRAPP complexes activate these GTPases via nucleotide exchange using a shared set of core subunits. The basal specificity of the TRAPP core is towards Rab1, yet the TRAPPII complex is specific for Rab11. A steric gating mechanism has been proposed to explain TRAPPII counterselection against Rab1. Here we present cryoEM structures of the 22-subunit TRAPPII complex from budding yeast, including a TRAPPII-Rab11 nucleotide exchange intermediate. The Trs130 subunit provides a ″leg″ that positions the active site distal to the membrane surface, and this leg is required for steric gating. The related TRAPPIII complex is unable to activate Rab11 due to a repulsive interaction, which TRAPPII surmounts using the Trs120 subunit as a ″lid″ to enclose the active site. TRAPPII also adopts an open conformation enabling Rab11 to access and exit from the active site chamber.

Spontaneous polarization effects on solid high harmonic generation in ferroelectric lithium niobate crystals

High-order harmonic generation (HHG) in ferroelectric lithium niobate (LiNbO$_{3}$) is investigated theoretically by solving the semi-conductor Bloch equations. Because of the spontaneous polarization, even-order harmonics are produced in the HHG spectra of the LiNbO$_{3}$ crystal driven by a monochromatic multi-cycle 3300-nm laser. Our numerical calculation shows that they are originated from the suppression of one half-optical cycle HHG process in each cycle of the driving field due to the spontaneous polarization. We also illustrate that the spontaneous polarization will increase the harmonic yield and extend the maximally attainable cutoff energy at the same time. We further report that the carrier-envelope phase dependence of HHG spectra changes from a minimum period of $\pi$ rad to 2$\pi$ rad when the laser polarization direction is parallel/anti-parallel to the spontaneous polarization direction in LiNbO$_{3}$ crystal. This is promising to be utilized as an isolated attosecond pulse (IAP) gating mechanism. Moreover, the two-color relative phase dependence of HHG in LiNbO$_{3}$ is also investigated and shows broken inversion-symmetry.

Unveiling the Gating Mechanism of CRAC Channel: A Computational Study

CRAC channel is ubiquitous and its importance in the regulation of the immune system is testified by the severe immunodeficiencies caused by its mutations. In this work we took advantage of the availability of open and closed structures of this channel to run for the first time simulations of the whole gating process reaching the relevant time-scale with an enhanced sampling technique, Targeted Molecular Dynamics. Our simulations highlighted a complex allosteric propagation of the conformational change from peripheral helices, where the activator STIM1 binds, to the central pore helices. In agreement with mutagenesis data, our simulations revealed the key role of residue H206 whose displacement creates an empty space behind the hydrophobic region of the pore, thus releasing a steric brake and allowing the opening of the channel. Conversely, the process of pore closing culminates with the formation of a bubble that occludes the pore even in the absence of steric block. This mechanism, known as “hydrophobic gating”, has been observed in an increasing number of biological ion channels and also in artificial nanopores. Our study therefore shows promise not only to better understand the molecular origin of diseases caused by disrupted calcium signaling, but also to clarify the mode of action of hydrophobically gated ion channels, possibly even suggesting strategies for the biomimetic design of synthetic nanopores.

Hyperactivation of the proteasome core particle in Caenorhabditis elegans protects against proteotoxic stress and extends lifespan

Many age-related diseases (ARDs) including virtually all neurodegenerative diseases (NDs) are characterized by the accumulation of proteins that are thought to significantly contribute to disease pathogenesis. One of the cell’s primary systems for the degradation of misfolded/damaged proteins is the Ubiquitin Proteasome System (UPS), and its impairment is implicated in essentially all NDs. Thus, upregulating this system to combat NDs has garnered a great deal of interest in recent years. Various animal models have focused on increasing the total proteasome levels, but thus far, none have focused on intrinsic activation of the proteasome itself. With this in mind, we constructed a, first to our knowledge, animal model that endogenously expresses a hyperactive open-gate proteasome in Caenorhabditis elegans (C. elegans). The gate-destabilizing mutation introduced into the nematode germline created a viable nematode population with substantially enhanced proteasomal peptidase and unstructured protein degradation activity. These CRISPR edited nematodes showed a significantly increased lifespan and substantial resistance to oxidative/proteotoxic stress with surprisingly mild consequential phenotypes. These results show that introducing a constitutively active proteasome into a multicellular organism is feasible and suggests targeting the proteasome gating mechanism as a valid approach for future ARD research efforts in mammals.

The Stop Signal Task for Measuring Behavioral Inhibition in Mice With Increased Sensitivity and High-Throughput Operation

Ceasing an ongoing motor response requires action cancelation. This is impaired in many pathologies such as attention deficit disorder and schizophrenia. Action cancelation is measured by the stop signal task that estimates how quickly a motor response can be stopped when it is already being executed. Apart from human studies, the stop signal task has been used to investigate neurobiological mechanisms of action cancelation overwhelmingly in rats and only rarely in mice, despite the need for a genetic model approach. Contributing factors to the limited number of mice studies may be the long and laborious training that is necessary and the requirement for a very loud (100 dB) stop signal. We overcame these limitations by employing a fully automated home-cage-based setup. We connected a home-cage to the operant box via a gating mechanism, that allowed individual ID chipped mice to start sessions voluntarily. Furthermore, we added a negative reinforcement consisting of a mild air puff with escape option to the protocol. This specifically improved baseline inhibition to 94% (from 84% with the conventional approach). To measure baseline inhibition the stop is signaled immediately with trial onset thus measuring action restraint rather than action cancelation ability. A high baseline allowed us to measure action cancelation ability with higher sensitivity. Furthermore, our setup allowed us to reduce the intensity of the acoustic stop signal from 100 to 70 dB. We constructed inhibition curves from stop trials with daily adjusted delays to estimate stop signal reaction times (SSRTs). SSRTs (median 88 ms) were lower than reported previously, which we attribute to the observed high baseline inhibition. Our automated training protocol reduced training time by 17% while also promoting minimal experimenter involvement. This sensitive and labor efficient stop signal task procedure should therefore facilitate the investigation of action cancelation pathologies in genetic mouse models.