Linn Sofie Sæther, Thor Ueland, Beathe Haatveit, Luigi Angelo Maglanoc, Attila Szabo, Srdjan Djurovic, Pål Aukrust, Daniel Roelfs, Christine Mohn, Monica Bettina Elkjaer Greenwood Ormerod, Trine Vik Lagerberg, Nils Eiel Steen, Ingrid Melle, Ole Andreas Andreassen, Torill Ueland <<<

A potential relationship between dysregulation of immune/inflammatory pathways and cognitive impairment has been suggested in severe mental illnesses (SMI), such as schizophrenia (SZ) and bipolar (BD) spectrum disorders. However, multivariate relationships between peripheral inflammatory/immune-related markers and cognitive domains are unclear, and many studies do not account for inter-individual variance in both cognitive functioning and inflammatory/immune status. This study aimed to investigate covariance patterns between inflammatory/immune-related markers and cognitive domains and further elucidate heterogeneity in a large SMI and healthy control (HC) cohort (SZ = 343, BD = 289, HC = 770). We applied canonical correlation analysis (CCA) to identify modes of maximum covariation between a comprehensive selection of cognitive domains and inflammatory/immune markers. We found that poor verbal learning and psychomotor processing speed was associated with higher levels of interleukin-18 system cytokines and beta defensin 2, reflecting enhanced activation of innate immunity, a pattern augmented in SMI compared to HC. Applying hierarchical clustering on covariance patterns identified by the CCA revealed a high cognition-low immune dysregulation subgroup with predominantly HC (24% SZ, 45% BD, 74% HC) and a low cognition-high immune dysregulation subgroup predominantly consisting of SMI patients (76% SZ, 55% BD, 26% HC). These subgroups differed in IQ, years of education, age, CRP, BMI (all groups), level of functioning, symptoms and defined daily dose (DDD) of antipsychotics (SMI cohort). Our findings suggest a link between cognitive impairment and innate immune dysregulation in a subset of individuals with severe mental illness. <<<

The proliferation of telco networks and mobile terminals brings the accumulation of tremendous amounts of measure report(MR) data at a rapid pace. The MR data is generated by mobile objects while connecting to data services and is stored in backend data centers. To geo-tag or localize such MR data is believed to have a profound effect on the analytics and optimizations of telco and traffic networks. However, MR records are of noisy and partial observations regarding to mobile objects' geo-locations and hence pose challenges to accurate telco data localization. There have been quite a few attempts. Single-point localization methods map a MR record to a location, but come out with limited accuracies due to the ignorance of spatiotemporal coherence of successive MR records. Recent efforts on sequential localization techniques alleviate this by mapping a sequence of MR records to a trajectory. However, existing solutions are often with assumptions on specific models, e.g., mobility and signal strength distributions, or priori knowledge on topology space, e.g., road networks, limiting the deployment in practice. To this end, we propose a data-driven framework to tackle the challenges in sequential telco localization. We solely use raw MR records and a public third-party GPS dataset for the learning of the correlations between mobile objects' locations and MR records, requiring no model assumptions and priori knowledge. To handle the data-intensive workloads during the learning process, we use materialized views for efficient online localization and light-weighted indexing techniques for periodical parameters tuning, in order to improve the efficiency and scalability. Results on real data show that our solution achieves 58.8 percent improvement in median localization errors compared with state-of-art sequential localization techniques that require hypothesis models and priori knowledge, making our solution superior in terms of effectiveness, efficiency, and employability. <<<

Most genetic variations associated with human complex traits are located in non-coding genomic regions. Therefore, understanding the genotype-to-phenotype axis requires a comprehensive catalog of functional non-coding genomic elements, most of which are involved in epigenetic regulation of gene expression. Genome-wide maps of open chromatin regions can facilitate functional analysis of cis- and trans-regulatory elements via their connections with trait-associated sequence variants. Currently, Assay for Transposase Accessible Chromatin with high-throughput sequencing (ATAC-seq) is considered the most accessible and cost-effective strategy for genome-wide profiling of chromatin accessibility. Single-cell ATAC-seq (scATAC-seq) technology has also been developed to study cell type-specific chromatin accessibility in tissue samples containing a heterogeneous cellular population. However, due to the intrinsic nature of scATAC-seq data, which are highly noisy and sparse, accurate extraction of biological signals and devising effective biological hypothesis are difficult. To overcome such limitations in scATAC-seq data analysis, new methods and software tools have been developed over the past few years. Nevertheless, there is no consensus for the best practice of scATAC-seq data analysis yet. In this review, we discuss scATAC-seq technology and data analysis methods, ranging from preprocessing to downstream analysis, along with an up-to-date list of published studies that involved the application of this method. We expect this review will provide a guideline for successful data generation and analysis methods using appropriate software tools and databases for the study of chromatin accessibility at single-cell resolution. <<<

Zachariah H Foda, Akshaya V Annapragada, Kavya Boyapati, Daniel C Bruhm, Nicholas A Vulpescu, Jamie E Medina, Dimitrios Mathios, Stephen Cristiano, Noushin Niknafs, Harry T Luu, Michael G Goggins, Robert A Anders, Jing Sun, Shruti H Meta, David L Thomas, Gregory D Kirk, Vilmos Adleff, Jillian Phallen, Robert B Scharpf, Amy K Kim, Victor E Velculescu <<<

Liver cancer is a major cause of cancer mortality worldwide. Screening individuals at high risk, including those with cirrhosis and viral hepatitis, provides an avenue for improved survival, but current screening methods are inadequate. In this study, we used whole-genome cell-free DNA (cfDNA) fragmentome analyses to evaluate 724 individuals from the United States, the European Union, or Hong Kong with hepatocellular carcinoma (HCC) or who were at average or high-risk for HCC. Using a machine learning model that incorporated multifeature fragmentome data, the sensitivity for detecting cancer was 88% in an average-risk population at 98% specificity and 85% among high-risk individuals at 80% specificity. We validated these results in an independent population. cfDNA fragmentation changes reflected genomic and chromatin changes in liver cancer, including from transcription factor binding sites. These findings provide a biological basis for changes in cfDNA fragmentation in patients with liver cancer and provide an accessible approach for noninvasive cancer detection. <<<

Manuela Costa, Diego Lozano-Soldevilla, Antonio Gil-Nagel, Rafael Toledano, Carina R Oehrn, Lukas Kunz, Mar Yebra, Costantino Mendez-Bertolo, Lennart Stieglitz, Johannes Sarnthein, Nikolai Axmacher, Stephan Moratti, Bryan A Strange <<<

Memory for aversive events is central to survival but can become maladaptive in psychiatric disorders. Memory enhancement for emotional events is thought to depend on amygdala modulation of hippocampal activity. However, the neural dynamics of amygdala-hippocampal communication during emotional memory encoding remain unknown. Using simultaneous intracranial recordings from both structures in human patients, here we show that successful emotional memory encoding depends on the amygdala theta phase to which hippocampal gamma activity and neuronal firing couple. The phase difference between subsequently remembered vs. not-remembered emotional stimuli translates to a time period that enables lagged coherence between amygdala and downstream hippocampal gamma. These results reveal a mechanism whereby amygdala theta phase coordinates transient amygdala -hippocampal gamma coherence to facilitate aversive memory encoding. Pacing of lagged gamma coherence via amygdala theta phase may represent a general mechanism through which the amygdala relays emotional content to distant brain regions to modulate other aspects of cognition, such as attention and decision-making. <<<

We present a novel framework for multivariate time series representation learning based on the transformer encoder architecture. The framework includes an unsupervised pre-training scheme, which can offer substantial performance benefits over fully supervised learning on downstream tasks, both with but even without leveraging additional unlabeled data, i.e., by reusing the existing data samples. Evaluating our framework on several public multivariate time series datasets from various domains and with diverse characteristics, we demonstrate that it performs significantly better than the best currently available methods for regression and classification, even for datasets which consist of only a few hundred training samples. Given the pronounced interest in unsupervised learning for nearly all domains in the sciences and in industry, these findings represent an important landmark, presenting the first unsupervised method shown to push the limits of state-of-the-art performance for multivariate time series regression and classification. <<<

Artists rarely paint in a single style throughout their career. More often they change styles or develop variations of it. In addition, artworks in different styles and even within one style depict real content differently: while Picasso's Blue Period displays a vase in a blueish tone but as a whole, his Cubist works deconstruct the object. To produce artistically convincing stylizations, style transfer models must be able to reflect these changes and variations. Recently many works have aimed to improve the style transfer task, but neglected to address the described observations. We present a novel approach which captures particularities of style and the variations within and separates style and content. This is achieved by introducing two novel losses: a fixpoint triplet style loss to learn subtle variations within one style or between different styles and a disentanglement loss to ensure that the stylization is not conditioned on the real input photo. In addition the paper proposes various evaluation methods to measure the importance of both losses on the validity, quality and variability of final stylizations. We provide qualitative results to demonstrate the performance of our approach. <<<

Genomic microarrays used to assess DNA copy number are now recommended as first-tier tests for the postnatal evaluation of individuals with intellectual disability, autism spectrum disorders, and/or multiple congenital anomalies. Application of this technology has resulted in the discovery of widespread copy number variation in the human genome, both polymorphic variation in healthy individuals and novel pathogenic copy number imbalances. To assist clinical laboratories in the evaluation of copy number variants and to promote consistency in interpretation and reporting of genomic microarray results, the American College of Medical Genetics has developed the following professional guidelines for the interpretation and reporting of copy number variation. These guidelines apply primarily to evaluation of constitutional copy number variants detected in the postnatal setting. <<<

AbstractRecent human electrophysiology work has uncovered the presence of high frequency oscillatory events, termed ripples, during awake behavior. This prior work focuses on ripples in the medial temporal lobe (MTL) during memory retrieval. Few studies, however, investigate ripples during item encoding. Many studies have found neural activity during encoding that predicts later recall, termed subsequent memory effects (SMEs), but it is unclear if ripples during encoding also predict subsequent recall. Detecting ripples in 124 neurosurgical participants performing an episodic memory task, we find insignificant ripple SMEs in any MTL region, even as these regions exhibit robust high frequency activity (HFA) SMEs. Instead, hippocampal ripples increase during encoding of items leading to recall of temporally or semantically associated items, a phenomenon known as clustering. This subsequent clustering effect (SCE) arises specifically when hippocampal ripples occur during both encoding and retrieval, suggesting that ripples mediate the encoding and future reinstatement of episodic memories. <<<

Development of ketosis in high-producing dairy cows contributes to several animal health issues and highlights the need for a better understanding of the genetic basis of metabolic diseases. To evaluate the pattern of differential gene expression in the liver of cows under negative energy balance (NEB), and under subclinical and clinical ketosis, a meta-analysis of gene expression and genome-wide association studies results was performed. An initial systematic review identified 118 articles based on the key words "cow," "liver," "negative energy balance," "ketosis," "expression," "qPCR," "microarray," "proteomic," "RNA-Seq," and "GWAS." After further screening for only peer-reviewed and pertinent articles for gene expression during NEB and clinical and subclinical ketosis (considering plasma levels of β-hydroxybutyrate), 20 articles were included in the analysis. From the systematic review, 430 significant SNPs identified by genome-wide association studies (GWAS) were assigned to genes reported in gene expression studies by considering chromosome and base pair positions in the ARS-UCD 1.2 bovine assembly. Venn diagrams were created to integrate the data obtained in the systematic review, and Gene Ontology enrichment analysis was carried out using official gene names. A QTL enrichment analysis was also performed to identify potential positional candidate loci. Twenty-four significant SNPs were located within the coordinates of differentially expressed genes located on chromosomes 2, 3, 6, 9, 11, 14, 27, and 29. Three significant metabolic pathways were associated with NEB and subclinical and clinical ketosis. In addition, 2 important genes, PPARA (peroxisome proliferator activated receptor alpha) and ACACA (acetyl-coenzyme A carboxylase α), were identified, which were differentially expressed in the 3 metabolic conditions. The PPARA gene is involved in the regulation of lipid metabolism and fatty liver disease and the ACACA gene encodes an enzyme that catalyzes the carboxylation of acetyl-coenzyme A to malonyl-coenzyme A, which is a rate-limiting step in fatty acid synthesis. Gene network analysis revealed co-expression interactions among 34 genes associated with functions involving fatty acid transport and fatty acid metabolism. For the annotated QTL, 9 QTL were identified for ketosis. The genes FN1 (fibronectin 1) and PTK2 (protein tyrosine kinase 2), which are mainly involved in cell adhesion and formation of extracellular matrix constituents, were enriched for QTL previously associated with the trait "ketosis" on chromosome 2 and for the trait "milk iron content" on chromosome 14, respectively. This integration of gene expression and GWAS data provides an additional understanding of the genetic background of NEB and subclinical and clinical ketosis in dairy cattle. Thus, it is a useful approach to identify biological mechanisms underlying these metabolic conditions in dairy cattle. <<<

S M Ashiqul Islam, Marcos Díaz-Gay, Yang Wu, Mark Barnes, Raviteja Vangara, Erik N Bergstrom, Yudou He, Mike Vella, Jingwei Wang, Jon W Teague, Peter Clapham, Sarah Moody, Sergey Senkin, Yun Rose Li, Laura Riva, Tongwu Zhang, Andreas J Gruber, Christopher D Steele, Burçak Otlu, Azhar Khandekar, Ammal Abbasi, Laura Humphreys, Natalia Syulyukina, Samuel W Brady, Boian S Alexandrov, Nischalan Pillay, Jinghui Zhang, David J Adams, Iñigo Martincorena, David C Wedge, Maria Teresa Landi, Paul Brennan, Michael R Stratton, Steven G Rozen, Ludmil B Alexandrov <<<

Mutational signature analysis is commonly performed in cancer genomic studies. Here, we present SigProfilerExtractor, an automated tool for extraction of mutational signatures, and benchmark it against another 13 bioinformatics tools by using 34 scenarios encompassing 2,500 simulated signatures found in 60,000 synthetic genomes and 20,000 synthetic exomes. For simulations with 5% noise, reflecting high-quality datasets, SigProfilerExtractor outperforms other approaches by elucidating between 20% and 50% more true-positive signatures while yielding 5-fold less false-positive signatures. Applying SigProfilerExtractor to 4,643 whole-genome- and 19,184 whole-exome-sequenced cancers reveals four novel signatures. Two of the signatures are confirmed in independent cohorts, and one of these signatures is associated with tobacco smoking. In summary, this report provides a reference tool for analysis of mutational signatures, a comprehensive benchmarking of bioinformatics tools for extracting signatures, and several novel mutational signatures, including one putatively attributed to direct tobacco smoking mutagenesis in bladder tissues. <<<

Lian Narunsky-Haziza , Gregory D. Sepich-Poore , Ilana Livyatan , Omer Asraf , Cameron Martino , Deborah Nejman , Nancy Gavert , Jason E. Stajich , Guy Amit , Antonio González , Stephen Wandro , Gili Perry , Ruthie Ariel , Arnon Meltser , Justin P. Shaffer , Qiyun Zhu , Nora Balint-Lahat , Iris Barshack , Maya Dadiani , Einav N. Gal-Yam , Sandip Pravin Patel , Amir Bashan , Austin D. Swafford , Yitzhak Pilpel , Rob Knight , Ravid Straussman <<<

Cancer-microbe associations have been explored for centuries, but cancer-associated fungi have rarely been examined. Here, we comprehensively characterize the cancer mycobiome within 17,401 patient tissue, blood, and plasma samples across 35 cancer types in four independent cohorts. We report fungal DNA and cells at low abundances across many major human cancers, with differences in community compositions that differ among cancer types, even when accounting for technical background. Fungal histological staining of tissue microarrays supported intratumoral presence and frequent spatial association with cancer cells and macrophages. Comparing intratumoral fungal communities with matched bacteriomes and immunomes revealed co-occurring bi-domain ecologies, often with permissive, rather than competitive, microenvironments and distinct immune responses. Clinically focused assessments suggested prognostic and diagnostic capacities of the tissue and plasma mycobiomes, even in stage I cancers, and synergistic predictive performance with bacteriomes. <<<

Prolonged detection of severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) RNA and recurrence of PCR-positive tests have been widely reported in patients after recovery from COVID-19, but some of these patients do not appear to shed infectious virus. We investigated the possibility that SARS-CoV-2 RNAs can be reverse-transcribed and integrated into the DNA of human cells in culture and that transcription of the integrated sequences might account for some of the positive PCR tests seen in patients. In support of this hypothesis, we found that DNA copies of SARS-CoV-2 sequences can be integrated into the genome of infected human cells. We found target site duplications flanking the viral sequences and consensus LINE1 endonuclease recognition sequences at the integration sites, consistent with a LINE1 retrotransposon-mediated, target-primed reverse transcription and retroposition mechanism. We also found, in some patient-derived tissues, evidence suggesting that a large fraction of the viral sequences is transcribed from integrated DNA copies of viral sequences, generating viral-host chimeric transcripts. The integration and transcription of viral sequences may thus contribute to the detection of viral RNA by PCR in patients after infection and clinical recovery. Because we have detected only subgenomic sequences derived mainly from the 3' end of the viral genome integrated into the DNA of the host cell, infectious virus cannot be produced from the integrated subgenomic SARS-CoV-2 sequences. <<<

Spatiotemporal (four-dimensional) infant-dedicated brain atlases are essential for neuroimaging analysis of early dynamic brain development. However, due to the substantial technical challenges in the acquisition and processing of infant brain MR images, 4D atlases densely covering the dynamic brain development during infancy are still scarce. Few existing ones generally have fuzzy tissue contrast and low spatiotemporal resolution, leading to degraded accuracy of atlas-based normalization and subsequent analyses. To address this issue, in this paper, we construct a 4D structural MRI atlas for infant brains based on the UNC/UMN Baby Connectome Project (BCP) dataset, which features a high spatial resolution, extensive age-range coverage, and densely sampled time points. Specifically, 542 longitudinal T1w and T2w scans from 240 typically developing infants up to 26-month of age were utilized for our atlas construction. To improve the co-registration accuracy of the infant brain images, which typically exhibit dynamic appearance with low tissue contrast, we employed the state-of-the-art registration method and leveraged our generated reliable brain tissue probability maps in addition to the intensity images to improve the alignment of individual images. To achieve consistent region labeling on both infant and adult brain images for facilitating region-based analysis across ages, we mapped the widely used Desikan cortical parcellation onto our atlas by following an age-decreasing mapping manner. Meanwhile, the typical subcortical structures were manually delineated to facilitate the studies related to the subcortex. Compared with the existing infant brain atlases, our 4D atlas has much higher spatiotemporal resolution and preserves more structural details, and thus can boost accuracy in neurodevelopmental analysis during infancy. <<<

Depression is a highly heterogeneous syndrome that bears only modest correlations with its biological substrates, motivating a renewed interest in rethinking our approach to diagnosing depression for research purposes and new efforts to discover subtypes of depression anchored in biology. Here, we review the major causes of diagnostic heterogeneity in depression, with consideration of both clinical symptoms and behaviors (symptomatology and trajectory of depressive episodes) and biology (genetics and sexually dimorphic factors). Next, we discuss the promise of using data-driven strategies to discover novel subtypes of depression based on functional neuroimaging measures, including dimensional, categorical, and hybrid approaches to parsing diagnostic heterogeneity and understanding its biological basis. The merits of using resting-state functional magnetic resonance imaging functional connectivity techniques for subtyping are considered along with a set of technical challenges and potential solutions. We conclude by identifying promising future directions for defining neurobiologically informed depression subtypes and leveraging them in the future for predicting treatment outcomes and informing clinical decision making. <<<

Christopher G Bell, Robert Lowe, Peter D Adams, Andrea A Baccarelli, Stephan Beck, Jordana T Bell, Brock C Christensen, Vadim N Gladyshev, Bastiaan T Heijmans, Steve Horvath, Trey Ideker, Jean-Pierre J Issa, Karl T Kelsey, Riccardo E Marioni, Wolf Reik, Caroline L Relton, Leonard C Schalkwyk, Andrew E Teschendorff, Wolfgang Wagner, Kang Zhang, Vardhman K Rakyan <<<

Epigenetic clocks comprise a set of CpG sites whose DNA methylation levels measure subject age. These clocks are acknowledged as a highly accurate molecular correlate of chronological age in humans and other vertebrates. Also, extensive research is aimed at their potential to quantify biological aging rates and test longevity or rejuvenating interventions. Here, we discuss key challenges to understand clock mechanisms and biomarker utility. This requires dissecting the drivers and regulators of age-related changes in single-cell, tissue- and disease-specific models, as well as exploring other epigenomic marks, longitudinal and diverse population studies, and non-human models. We also highlight important ethical issues in forensic age determination and predicting the trajectory of biological aging in an individual. <<<

Handwritten Chinese text recognition (HCTR) has been an active research topic for decades. However, most previous studies solely focus on the recognition of cropped text line images, ignoring the error caused by text line detection in real-world applications. Although some approaches aimed at page-level text recognition have been proposed in recent years, they either are limited to simple layouts or require very detailed annotations including expensive line-level and even character-level bounding boxes. To this end, we propose PageNet for end-to-end weakly supervised page-level HCTR. PageNet detects and recognizes characters and predicts the reading order between them, which is more robust and flexible when dealing with complex layouts including multi-directional and curved text lines. Utilizing the proposed weakly supervised learning framework, PageNet requires only transcripts to be annotated for real data; however, it can still output detection and recognition results at both the character and line levels, avoiding the labor and cost of labeling bounding boxes of characters and text lines. Extensive experiments conducted on five datasets demonstrate the superiority of PageNet over existing weakly supervised and fully supervised page-level methods. These experimental results may spark further research beyond the realms of existing methods based on connectionist temporal classification or attention. The source code is available at https://github.com/shannanyinxiang/PageNet. <<<

Katharine C Simon, Gregory D Clemenson, Jing Zhang, Negin Sattari, Alessandra E Shuster, Brandon Clayton, Elisabet Alzueta, Teji Dulai, Massimiliano de Zambotti, Craig Stark, Fiona C Baker, Sara C Mednick <<<

Sleep facilitates hippocampal-dependent memories, supporting the acquisition and maintenance of internal representation of spatial relations within an environment. In humans, however, findings have been mixed regarding sleep's contribution to spatial memory and navigation, which may be due to task designs or outcome measurements. We developed the Minecraft Memory and Navigation (MMN) task for the purpose of disentangling how spatial memory accuracy and navigation change over time, and to study sleep's independent contributions to each. In the MMN task, participants learned the locations of objects through free exploration of an open field computerized environment. At test, they were teleported to random positions around the environment and required to navigate to the remembered location of each object. In study 1, we developed and validated four unique MMN environments with the goal of equating baseline learning and immediate test performance. A total of 86 participants were administered the training phases and immediate test. Participants' baseline performance was equivalent across all four environments, supporting the use of the MMN task. In study 2, 29 participants were trained, tested immediately, and again 12 h later after a period of sleep or wake. We found that the metric accuracy of object locations, i.e., spatial memory, was maintained over a night of sleep, while after wake, metric accuracy declined. In contrast, spatial navigation improved over both sleep and wake delays. Our findings support the role of sleep in retaining the precise spatial relationships within a cognitive map; however, they do not support a specific role of sleep in navigation. <<<

Yan Li, Yang Chen, Lu Zhou, Shengjie You, Heng Deng, Ya Chen, Saleh Alseekh, Yong Yuan, Rao Fu, Zixin Zhang, Dan Su, Alisdair R Fernie, Mondher Bouzayen, Tao Ma, Mingchun Liu, Yang Zhang <<<

Tomato (Solanum lycopersicum) is a major horticultural crop worldwide and has emerged as a preeminent model for metabolic research. Although many research efforts have focused on the analysis of metabolite differences between varieties and species, the dynamics of metabolic changes during the tomato growth cycle and the regulatory networks that underlie these changes are poorly understood. In this study, we integrated high-resolution spatio-temporal metabolome and transcriptome data to systematically explore the metabolic landscape across 20 major tomato tissues and growth stages. In the resulting MicroTom Metabolic Network, the 540 detected metabolites and their co-expressed genes could be divided into 10 distinct clusters based on their biological functions. Using this dataset, we constructed a global map of the major metabolic changes that occur throughout the tomato growth cycle and dissected the underlying regulatory network. In addition to verifying previously well-established regulatory networks for important metabolites, we identified novel transcription factors that regulate the biosynthesis of important secondary metabolites such as steroidal glycoalkaloids and flavonoids. Our findings provide insights into spatio-temporal changes in tomato metabolism and generate a valuable resource for the study of metabolic regulatory processes in model plants. <<<

Structure-based drug design uses three-dimensional geometric information of macromolecules, such as proteins or nucleic acids, to identify suitable ligands. Geometric deep learning, an emerging concept of neural-network-based machine learning, has been applied to macromolecular structures. This review provides an overview of the recent applications of geometric deep learning in bioorganic and medicinal chemistry, highlighting its potential for structure-based drug discovery and design. Emphasis is placed on molecular property prediction, ligand binding site and pose prediction, and structure-based de novo molecular design. The current challenges and opportunities are highlighted, and a forecast of the future of geometric deep learning for drug discovery is presented. <<<