Shuyang Yao, Arvid Harder, Fahimeh Darki, Yu-Wei Chang, Ang Li, Kasra Nikouei, Giovanni Volpe, Johan N Lundström, Jian Zeng, Naomi Wray, Yi Lu, Patrick F Sullivan, Jens Hjerling-Leffler <<<

AbstractUnderstanding the temporal and spatial brain locations etiological for psychiatric disorders is essential for targeted neurobiological research. Integration of genomic insights from genome-wide association studies with single-cell transcriptomics is a powerful approach although past efforts have necessarily relied on mouse atlases. Leveraging a comprehensive atlas of the adult human brain, we prioritized cell types via the enrichment of SNP-heritabilities for brain diseases, disorders, and traits, progressing from individual cell types to brain regions. Our findings highlight specific neuronal clusters significantly enriched for the SNP-heritabilities for schizophrenia, bipolar disorder, and major depressive disorder along with intelligence, education, and neuroticism. Extrapolation of cell-type results to brain regions reveals important patterns for schizophrenia with distinct subregions in the hippocampus and amygdala exhibiting the highest significance. Cerebral cortical regions display similar enrichments despite the known prefrontal dysfunction in those with schizophrenia highlighting the importance of subcortical connectivity. Using functional MRI connectivity from cases with schizophrenia and neurotypical controls, we identified brain networks that distinguished cases from controls that also confirmed involvement of the central and lateral amygdala, hippocampal body, and prefrontal cortex. Our findings underscore the value of single-cell transcriptomics in decoding the polygenicity of psychiatric disorders and offer a promising convergence of genomic, transcriptomic, and brain imaging modalities toward common biological targets. <<<

Elena Beanato, Hyuk-June Moon, Fabienne Windel, Pierre Vassiliadis, Maximillian J. Wessel, Traian Popa, Menoud Pauline, Esra Neufeld, Emanuela De Falco, Baptiste Gauthier, Melanie Steiner, Olaf Blanke, Friedhelm C. Hummel <<<

Because of the depth of the hippocampal-entorhinal complex (HC-EC) in the brain, understanding of its role in spatial navigation via neuromodulation was limited in humans. Here, we aimed to better elucidate this relationship in healthy volunteers, using transcranial temporal interference electric stimulation (tTIS), a noninvasive technique allowing to selectively neuromodulate deep brain structures. We applied tTIS to the right HC-EC in either continuous or intermittent theta-burst stimulation patterns (cTBS or iTBS), compared to a control condition, during a virtual reality–based spatial navigation task and concomitant functional magnetic resonance imaging. iTBS improved spatial navigation performance, correlated with hippocampal activity modulation, and decreased grid cell–like activity in EC. Collectively, these data provide the evidence that human HC-EC activity can be directly and noninvasively modulated leading to changes of spatial navigation behavior. These findings suggest promising perspectives for patients suffering from cognitive impairment such as following traumatic brain injury or dementia. <<<

AbstractThe hippocampal-entorhinal system uses cognitive maps to represent spatial knowledge and other types of relational information, such as the transition probabilities between objects. However, objects can often be characterized in terms of different types of relations simultaneously, e.g. semantic similarities learned over the course of a lifetime as well as transitions experienced over a brief timeframe in an experimental setting. Here we ask how the hippocampal formation handles the embedding of stimuli in multiple relational structures that differ vastly in terms of their mode and timescale of acquisition: Does it integrate the different stimulus dimensions into one conjunctive map, or is each dimension represented in a parallel map? To this end, we reanalyzed functional magnetic resonance imaging (fMRI) data from Garvert et al. (2017) that had previously revealed an entorhinal map which coded for newly learnt statistical regularities. We used a triplet odd-one-out task to construct a semantic distance matrix for presented items and applied fMRI adaptation analysis to show that the degree of similarity of representations in bilateral hippocampus decreases as a function of semantic distance between presented objects. Importantly, while both maps localize to the hippocampal formation, this semantic map is anatomically distinct from the originally described entorhinal map. This finding supports the idea that the hippocampal-entorhinal system forms parallel cognitive maps reflecting the embedding of objects in diverse relational structures. <<<

Noninvasive prenatal testing (NIPT) for the sex chromosome aneuploidies (45,X, 47,XXY, 47,XXX, and 47,XYY) differs significantly from that for the autosomal aneuploidies (trisomy 13, 18, and 21). As a group, sex chromosome aneuploidies occur more commonly (1/400) than any one isolated autosomal aneuploidy, the phenotypic variation is greater, the role of mosaicism more challenging, and the positive predictive value of a high-risk NIPT result is substantially lower. These considerations should be identified during pretest counseling, the inclusion of sex chromosome testing offered separately, and the differences from autosomal aneuploidy NIPT clearly delineated. <<<

Nga Yan Tse, Aswin Ratheesh, Ye Ella Tian, Colm G. Connolly, Christopher G. Davey, Saampras Ganesan, Ian H. Gotlib, Ben J. Harrison, Laura K. M. Han, Tiffany C. Ho, Alec J. Jamieson, Jaclyn S. Kirshenbaum, Yong Liu, Xiaohong Ma, Amar Ojha, Jiang Qiu, Matthew D. Sacchet, Lianne Schmaal, Alan N. Simmons, John Suckling, Dongtao Wei, Xiao Yang, Tony T. Yang, Robin F. H. Cash, Andrew Zalesky <<<

Classical optimization and learning-based methods are the two reigningparadigms in deformable image registration. While optimization-based methodsboast generalizability across modalities and robust performance, learning-basedmethods promise peak performance, incorporating weak supervision and amortizedoptimization. However, the exact conditions for either paradigm to perform wellover the other are shrouded and not explicitly outlined in the existingliterature. In this paper, we make an explicit correspondence between themutual information of the distribution of per-pixel intensity and labels, andthe performance of classical registration methods. This strong correlationhints to the fact that architectural designs in learning-based methods isunlikely to affect this correlation, and therefore, the performance oflearning-based methods. This hypothesis is thoroughly validated withstate-of-the-art classical and learning-based methods. However, learning-basedmethods with weak supervision can perform high-fidelity intensity and labelregistration, which is not possible with classical methods. Next, we show thatthis high-fidelity feature learning does not translate to invariance to domainshift, and learning-based methods are sensitive to such changes in the datadistribution. Finally, we propose a general recipe to choose the best paradigmfor a given registration problem, based on these observations. <<<

Zhaolian Lu, Shanlan Mo, Duo Xie, Xiangwei Zhai, Shanjun Deng, Kantian Zhou, Kun Wang, Xueling Kang, Hao Zhang, Juanzhen Tong, Liangzhen Hou, Huijuan Hu, Xuefei Li, Da Zhou, Leo Tsz On Lee, Li Liu, Yaxi Zhu, Jing Yu, Ping Lan, Jiguang Wang, Zhen He, Xionglei He, Zheng Hu <<<

In the last ten years, different concepts of electric vertical take-off and landing aircrafts (eVTOLs) have been tested. This article addresses the problem of the choice of the best configuration. VTOLs built since the fifties are presented and their advantages, disadvantages, and problems are discussed. Three representative eVTOLs, one for each main configuration, are compared on five main parameters and three reference missions. The parameters are disk loading, total hover time, cruise speed, practical range, and flight time. The performance of the eVTOLs on the urban, extra-urban, and long-range mission is evaluated computing the time and energy required. The results show that the best configuration depends on the mission. The multirotor is more efficient in hover. The vectored thrust jet is more efficient in cruise and has a higher range. The lift + cruise is a compromise. <<<

Rasmus Siersbæk, Jesper Grud Skat Madsen, Biola Maria Javierre, Ronni Nielsen, Emilie Kristine Bagge, Jonathan Cairns, Steven William Wingett, Sofie Traynor, Mikhail Spivakov, Peter Fraser, Susanne Mandrup <<<

Abstract Inflorescence branching in the grasses controls the number of florets and hence the number of seeds. Recent data on the underlying genetics come primarily from rice and maize, although new data are accumulating in other systems as well. This review focuses on a window in developmental time from the production of primary branches by the inflorescence meristem through to the production of glumes, which indicate the transition to producing a spikelet. Several major developmental regulatory modules appear to be conserved among most or all grasses. Placement and development of primary branches are controlled by conserved auxin regulatory genes. Subtending bracts are repressed by a network including TASSELSHEATH4, and axillary branch meristems are regulated largely by signaling centers that are adjacent to but not within the meristems themselves. Gradients of SQUAMOSA-PROMOTER BINDING-like and APETALA2-like proteins and their microRNA regulators extend along the inflorescence axis and the branches, governing the transition from production of branches to production of spikelets. The relative speed of this transition determines the extent of secondary and higher order branching. This inflorescence regulatory network is modified within individual species, particularly as regards formation of secondary branches. Differences between species are caused both by modifications of gene expression and regulators and by presence or absence of critical genes. The unified networks described here may provide tools for investigating orphan crops and grasses other than the well-studied maize and rice. <<<

Sequential fine-tuning and multi-task learning are methods aiming toincorporate knowledge from multiple tasks; however, they suffer fromcatastrophic forgetting and difficulties in dataset balancing. To address theseshortcomings, we propose AdapterFusion, a new two stage learning algorithm thatleverages knowledge from multiple tasks. First, in the knowledge extractionstage we learn task specific parameters called adapters, that encapsulate thetask-specific information. We then combine the adapters in a separate knowledgecomposition step. We show that by separating the two stages, i.e., knowledgeextraction and knowledge composition, the classifier can effectively exploitthe representations learned from multiple tasks in a non-destructive manner. Weempirically evaluate AdapterFusion on 16 diverse NLU tasks, and find that iteffectively combines various types of knowledge at different layers of themodel. We show that our approach outperforms traditional strategies such asfull fine-tuning as well as multi-task learning. Our code and adapters areavailable at AdapterHub.ml. <<<

Ivan Anishchenko, Yakov Kipnis, Indrek Kalvet, Guangfeng Zhou, Rohith Krishna, Samuel J. Pellock, Anna Lauko, Gyu Rie Lee, Linna An, Justas Dauparas, Frank DiMaio, David Baker <<<

AbstractModeling the conformational heterogeneity of protein-small molecule systems is an outstanding challenge. We reasoned that while residue level descriptions of biomolecules are efficient for de novo structure prediction, for probing heterogeneity of interactions with small molecules in the folded state an entirely atomic level description could have advantages in speed and generality. We developed a graph neural network called ChemNet trained to recapitulate correct atomic positions from partially corrupted input structures from the Cambridge Structural Database and the Protein Data Bank; the nodes of the graph are the atoms in the system. ChemNet accurately generates structures of diverse organic small molecules given knowledge of their atom composition and bonding, and given a description of the larger protein context, and builds up structures of small molecules and protein side chains for protein-small molecule docking. Because ChemNet is rapid and stochastic, ensembles of predictions can be readily generated to map conformational heterogeneity. In enzyme design efforts described here and elsewhere, we find that using ChemNet to assess the accuracy and pre-organization of the designed active sites results in higher success rates and higher activities; we obtain a preorganized retroaldolase with akcat/KMof 11000 M-1min- 1, considerably higher than any pre-deep learning design for this reaction. We anticipate that ChemNet will be widely useful for rapidly generating conformational ensembles of small molecule and small molecule-protein systems, and for designing higher activity preorganized enzymes. <<<

Kimberly Siletti, Rebecca Hodge, Alejandro Mossi Albiach, Ka Wai Lee, Song-Lin Ding, Lijuan Hu, Peter Lönnerberg, Trygve Bakken, Tamara Casper, Michael Clark, Nick Dee, Jessica Gloe, Daniel Hirschstein, Nadiya V. Shapovalova, C. Dirk Keene, Julie Nyhus, Herman Tung, Anna Marie Yanny, Ernest Arenas, Ed S. Lein, Sten Linnarsson <<<

The human brain directs complex behaviors, ranging from fine motor skills to abstract intelligence, but the diversity of cell types that support these skills has not been fully described. In this work, we used single-nucleus RNA sequencing to systematically survey cells across the entire adult human brain. We sampled more than three million nuclei from approximately 100 dissections across the forebrain, midbrain, and hindbrain in three postmortem donors. Our analysis identified 461 clusters and 3313 subclusters organized largely according to developmental origins and revealing high diversity in midbrain and hindbrain neurons. Astrocytes and oligodendrocyte-lineage cells also exhibited regional diversity at multiple scales. The transcriptomic census of the entire human brain presented in this work provides a resource for understanding the molecular diversity of the human brain in health and disease. <<<

Transformer models have demonstrated impressive capabilities when trained atscale, excelling at difficult cognitive tasks requiring complex reasoning andrational decision-making. In this paper, we explore the application oftransformer models to chess, focusing on the critical role of the positionencoding within the attention mechanism. We show that in chess, transformersendowed with a sufficiently versatile position encoding can match existingchess-playing models at a fraction of the computational cost. Our architecturesignificantly outperforms AlphaZero at 8x fewer FLOPS and matches priorgrandmaster-level transformer-based agents at 30x fewer FLOPS. <<<