Search

Hypergraphs are instrumental in modeling complex relational systems that encompass a wide spectrum of high-order interactions among components. One prevalent analysis task is the properties estimation of large-scale hypergraphs, which involves selecting a subset of nodes and hyperedges while preserving the characteristics of the entire hypergraph. This paper aims to sample hypergraphs via random walks and is the first to perform unbiased random walks for sampling of nodes and hyperedges simultaneously in large-scale hypergraphs to the best of our knowledge. Initially, we analyze the stationary distributions of nodes and hyperedges for the simple random walk, and show that there is a high bias in both nodes and hyperedges. Subsequently, to eliminate the high bias of the simple random walk, we propose unbiased random walk strategies for nodes and hyperedges, respectively. Finally, a single joint walk schema is developed for sampling nodes and hyperedges simultaneously. To accelerate the convergence process, we employ delayed acceptance and history-aware techniques to assist our algorithm in achieving fast convergence. Extensive experimental results validate our theoretical findings, and the unbiased sampling algorithms for nodes and hyperedges have their complex hypergraph scenarios for which they are applicable. The joint random walk algorithm balanced the sampling applicable to both nodes and hyperedges. [ABSTRACT FROM AUTHOR]

Introduction: The visual signals evoked at the retinal ganglion cells are modified and modulated by various synaptic inputs that impinge on lateral geniculate nucleus cells before they are sent to the cortex. The selectivity of geniculate inputs for clustering or forming microcircuits on discrete dendritic segments of geniculate cell types may provide the structural basis for network properties of the geniculate circuitry and differential signal processing through the parallel pathways of vision. In our study, we aimed to reveal the patterns of input selectivity on morphologically discernable relay cell types and interneurons in the mouse lateral geniculate nucleus. Methods: We used two sets of Scanning Blockface Electron Microscopy (SBEM) image stacks and Reconstruct software to manually reconstruct of terminal boutons and dendrite segments. First, using an unbiased terminal sampling (UTS) approach and statistical modeling, we identified the criteria for volumebased sorting of geniculate boutons into their putative origins. Geniculate terminal boutons that were sorted in retinal and non-retinal categories based on previously described mitochondrial morphology, could further be sorted into multiple subpopulations based on their bouton volume distributions. Terminals deemed non-retinal based on the morphological criteria consisted of five distinct subpopulations, including small-sized putative corticothalamic and cholinergic boutons, two medium-sized putative GABAergic inputs, and a large-sized bouton type that contains dark mitochondria. Retinal terminals also consisted of four distinct subpopulations. The cutoff criteria for these subpopulations were then applied to datasets of terminals that synapse on reconstructed dendrite segments of relay cells or interneurons. Results: Using a network analysis approach, we found an almost complete segregation of retinal and cortical terminals on putative X-type cell dendrite segments characterized by grape-like appendages and triads. On these cells, interneuron appendages intermingle with retinal and other medium size terminals to form triads within glomeruli. In contrast, a second, presumed Y-type cell displayed dendrodendritic puncta adherentia and received all terminal types without a selectivity for synapse location; these were not engaged in triads. Furthermore, the contribution of retinal and cortical synapses received by X-, Yand interneuron dendrites differed such that over 60% of inputs to interneuron dendrites were from the retina, as opposed to 20% and 7% to X- and Y-type cells, respectively. Conclusion: The results underlie differences in network properties of synaptic inputs from distinct origins on geniculate cell types. [ABSTRACT FROM AUTHOR]

IntroductionThe visual signals evoked at the retinal ganglion cells are modified and modulated by various synaptic inputs that impinge on lateral geniculate nucleus cells before they are sent to the cortex. The selectivity of geniculate inputs for clustering or forming microcircuits on discrete dendritic segments of geniculate cell types may provide the structural basis for network properties of the geniculate circuitry and differential signal processing through the parallel pathways of vision. In our study, we aimed to reveal the patterns of input selectivity on morphologically discernable relay cell types and interneurons in the mouse lateral geniculate nucleus.MethodsWe used two sets of Scanning Blockface Electron Microscopy (SBEM) image stacks and Reconstruct software to manually reconstruct of terminal boutons and dendrite segments. First, using an unbiased terminal sampling (UTS) approach and statistical modeling, we identified the criteria for volume-based sorting of geniculate boutons into their putative origins. Geniculate terminal boutons that were sorted in retinal and non-retinal categories based on previously described mitochondrial morphology, could further be sorted into multiple subpopulations based on their bouton volume distributions. Terminals deemed non-retinal based on the morphological criteria consisted of five distinct subpopulations, including small-sized putative corticothalamic and cholinergic boutons, two medium-sized putative GABAergic inputs, and a large-sized bouton type that contains dark mitochondria. Retinal terminals also consisted of four distinct subpopulations. The cutoff criteria for these subpopulations were then applied to datasets of terminals that synapse on reconstructed dendrite segments of relay cells or interneurons.ResultsUsing a network analysis approach, we found an almost complete segregation of retinal and cortical terminals on putative X-type cell dendrite segments characterized by grape-like appendages and triads. On these cells, interneuron appendages intermingle with retinal and other medium size terminals to form triads within glomeruli. In contrast, a second, presumed Y-type cell displayed dendrodendritic puncta adherentia and received all terminal types without a selectivity for synapse location; these were not engaged in triads. Furthermore, the contribution of retinal and cortical synapses received by X-, Y- and interneuron dendrites differed such that over 60% of inputs to interneuron dendrites were from the retina, as opposed to 20% and 7% to X- and Y-type cells, respectively.ConclusionThe results underlie differences in network properties of synaptic inputs from distinct origins on geniculate cell types.

Day-fresh Great Cormorant pellets were collected at a huge single winter roost at Kreupel island in Lake IJsselmeer, The Netherlands. The roost is situated on a sandy island, 3 m above the water level and is regularly used by 1000–8000+ Cormorants. By examining pellets and thus reconstructing the previous day's meal of Cormorants gathering at the roost, we explored spatio-temporal differences in the diet of individual birds. We found that structural differences in pellets (size, colour and structure) are related to their content (fish species and fresh mass of a daily meal). Structurally larger and lighter coloured pellets contained larger and coarser fish and represented a higher total fish mass. Pellets of different size and structure were non-randomly distributed at the roost. Birds positioned centrally in the roost produced on average larger pellets than birds at the edges. Cormorants having few or no food remains in their pellets were more frequently recorded on the edge of the roost than in the centre and in particular, they were more likely to be located to the rear of the roost. The same was true for pellets showing a bloody signature which may be linked to either injury by sharp fin rays or to a parasite induced effect. Pellet data were corroborated by observations of differential distribution of faecal splashes of Cormorants, showing that individuals in the centre of the roost produced most guano. Clear diet differences existed within the roost between birds in the centre and elsewhere in the roost, whereas smaller differences existed between the front, edge and rear of the roost. Our data suggest a social hierarchy based on higher foraging success of birds returning earlier to the roost and settling in the centre, hereby forcing later arriving and less successful Cormorants towards the edge, rear and front position. We conclude that position at the roost is a reflection of previous foraging success, possibly caused by differences in dominance between Cormorants during the period of communal foraging. From a methodological point of view, our study demonstrates that, because of structural differences in pellets and spatial differences in occurrence, it is important to collect pellets randomly if one is to describe a proper diet sample by using pellets at any given site. Especially in marshlands and under trees and bush growth with a lot of litter, the likely collection of larger, more conspicuous pellets will cause a bias towards the larger fish species and sizes. Also sampling only at edge locations or specifically at the centre of a roost may give biased results, as this may result in an under- or overestimation in reconstructed fish uptake (e.g. fish size, mass, species) from the waterbody under study. [ABSTRACT FROM AUTHOR]