Search

MotivationInteractions between proteins help us understand how genes are functionally related and how they contribute to phenotypes. Experiments provide imperfect ‘ground truth’ information about a small subset of potential interactions in a specific biological context, which can then be extended to the whole genome across different contexts, such as conditions, tissues or species, through machine learning methods. However, evaluating the performance of these methods remains a critical challenge. Here, we propose to evaluate the generalizability of gene characterizations through the shape of performance curves.ResultsWe identify Functional Equivalence Classes (FECs), subsets of annotated and unannotated genes that jointly drive performance, by assessing the presence of straight lines in ROC curves built from gene-centric prediction tasks, such as function or interaction predictions. FECs are widespread across data types and methods, they can be used to evaluate the extent and context-specificity of functional annotations in a data-driven manner. For example, FECs suggest that B cell markers can be decomposed into shared primary markers (10–50 genes), and tissue-specific secondary markers (100–500 genes). In addition, FECs suggest the existence of functional modules that span a wide range of the genome, with marker sets spanning at most 5% of the genome and data-driven extensions of Gene Ontology sets spanning up to 40% of the genome. Simple to assess visually and statistically, the identification of FECs in performance curves paves the way for novel functional characterization and increased robustness in the definition of functional gene sets.Availability and implementationCode for analyses and figures is available at https://github.com/yexilein/pyroc.Supplementary informationSupplementary data are available at Bioinformatics online.

Comparison of IGO- and OGO-derived lesions

Pancreatic ductal adenocarcinoma (PDAC) is the most lethal common malignancy, with little improvement in patient outcomes over the past decades. Recently, subtypes of pancreatic cancer with different prognoses have been elaborated; however, the inability to model these subtypes has precluded mechanistic investigation of their origins. Here, we present a xenotransplantation model of PDAC in which neoplasms originate from patient-derived organoids injected directly into murine pancreatic ducts. Our model enables distinction of the two main PDAC subtypes: intraepithelial neoplasms from this model progress in an indolent or invasive manner representing the classical or basal-like subtypes of PDAC, respectively. Parameters that influence PDAC subtype specification in this intraductal model include cell plasticity and hyperactivation of the RAS pathway. Finally, through intratumoral dissection and the direct manipulation of RAS gene dosage, we identify a suite of RAS-regulated secreted and membrane-bound proteins that may represent potential candidates for therapeutic intervention in patients with PDAC.Significance:Accurate modeling of the molecular subtypes of pancreatic cancer is crucial to facilitate the generation of effective therapies. We report the development of an intraductal organoid transplantation model of pancreatic cancer that models the progressive switching of subtypes, and identify stochastic and RAS-driven mechanisms that determine subtype specification.See related commentary by Pickering and Morton, p. 1448.This article is highlighted in the In This Issue feature, p. 1426

Confocal z-stack imaging of immunofluorescent (IF) images of mStrawberry-hT3 grafts 4 weeks after IGO transplantation

Supplementary Table S1 Survival times of IGO and OGO mice Supplementary Table S2 Characteristics of patient-derived organoids, including KRAS, TP53, SMAD4, CDKN2A mutation status and patient stage Supplementary Table S3 Summary of Survival Data, including engraftment rate, mean survival, metastatic frequency Supplementary Table S4 Primers for quantitative PCR

Method Used to Define the Thresholds for Fast- and Slow-Progressing IGO-Derived Tumors

The Human Reference Genome serves as the foundation for modern genomic analyses. However, in its present form, it does not adequately represent the vast genetic diversity of the human population. In this study, we explored the consensus genome as a potential successor of the current reference genome and assessed its effect on the accuracy of RNA-seq read alignment. To find the best haploid genome representation, we constructed consensus genomes at the pan-human, superpopulation, and population levels, using variant information from The 1000 Genomes Project Consortium. Using personal haploid genomes as the ground truth, we compared mapping errors for real RNA-seq reads aligned to the consensus genomes versus the reference genome. For reads overlapping homozygous variants, we found that the mapping error decreased by a factor of approximately two to three when the reference was replaced with the pan-human consensus genome. We also found that using more population-specific consensuses resulted in little to no increase over using the pan-human consensus, suggesting a limit in the utility of incorporating a more specific genomic variation. Replacing the reference with consensus genomes impacts functional analyses, such as differential expressions of isoforms, genes, and splice junctions.

Understanding how genetic variants impact molecular phenotypes is a key goal of functional genomics, currently hindered by reliance on a single haploid reference genome. Here, we present the EN-TEx resource of 1,635 open-access datasets from four donors (∼30 tissues × ∼15 assays). The datasets are mapped to matched, diploid genomes with long-read phasing and structural variants, instantiating a catalog of >1 million allele-specific loci. These loci exhibit coordinated activity along haplotypes and are less conserved than corresponding, non-allele-specific ones. Surprisingly, a deep-learning transformer model can predict the allele-specific activity based only on local nucleotide-sequence context, highlighting the importance of transcription-factor-binding motifs particularly sensitive to variants. Furthermore, combining EN-TEx with existing genome annotations reveals strong associations between allele-specific and GWAS loci. It also enables models for transferring known eQTLs to difficult-to-profile tissues (e.g., from skin to heart). Overall, EN-TEx provides rich data and generalizable models for more accurate personal functional genomics.

An essential step toward understanding brain function is to establish a structural framework with cellular resolution on which multi-scale datasets spanning molecules, cells, circuits and systems can be integrated and interpreted1. Here, as part of the collaborative Brain Initiative Cell Census Network (BICCN), we derive a comprehensive cell type-based anatomical description of one exemplar brain structure, the mouse primary motor cortex, upper limb area (MOp-ul). Using genetic and viral labelling, barcoded anatomy resolved by sequencing, single-neuron reconstruction, whole-brain imaging and cloud-based neuroinformatics tools, we delineated the MOp-ul in 3D and refined its sublaminar organization. We defined around two dozen projection neuron types in the MOp-ul and derived an input–output wiring diagram, which will facilitate future analyses of motor control circuitry across molecular, cellular and system levels. This work provides a roadmap towards a comprehensive cellular-resolution description of mammalian brain architecture., Multi-modal analysis is used to generate a 3D atlas of the upper limb area of the mouse primary motor cortex, providing a framework for future studies of motor control circuitry.

The cortex is composed of neuronal types with diverse gene expression that are organized into specialized cortical areas. These areas, each with characteristic cytoarchitecture1–3, connectivity4,5, and neuronal activity6–10, are wired into modular networks4,5,11. However, it remains unclear whether cortical areas and their modular organization can be similarly defined by their transcriptomic signatures. Here we used BARseq, a high-throughput in situ sequencing technique, to interrogate the expression of 107 cell type marker genes in 1.2 million cells over a mouse forebrain hemisphere at cellular resolution.De novoclustering of gene expression in single neurons revealed transcriptomic types that were consistent with previous single-cell RNAseq studies12,13. Within medium-grained cell types that are shared across all cortical areas, gene expression and the distribution of fine-grained cell types vary along the contours of cortical areas. The compositions of transcriptomic types are highly predictive of cortical area identity. We grouped cortical areas into modules so that areas within a module, but not across modules, had similar compositions of transcriptomic types. Strikingly, these modules match cortical subnetworks that are highly interconnected4,5,11, suggesting that cortical areas that are similar in cell types are also wired together. This “wire-by-similarity” rule reflects a novel organizing principle for the connectivity of cortical areas. Our BARseq-based strategy is high-throughput and low-cost, and scaling up this approach to many animals can potentially reveal the brain-wide molecular architecture across individuals, developmental times, and disease models.

Characterizing cellular diversity at different levels of biological organization across data modalities is a prerequisite to understanding the function of cell types in the brain. Classification of neurons is also required to manipulate cell types in controlled ways, and to understand their variation and vulnerability in brain disorders. TheBRAIN Initiative Cell Census Network (BICCN)is an integrated network of data generating centers, data archives and data standards developers, with the goal of systematic multimodal brain cell type profiling and characterization. Emphasis of the BICCN is on the whole mouse brain and demonstration of prototypes for human and non-human primate (NHP) brains. Here, we provide a guide to the cellular and spatial approaches employed, and to accessing and using the BICCN data and its extensive resources, including theBRAIN Cell Data Center (BCDC)which serves to manage and integrate data across the ecosystem. We illustrate the power of the BICCN data ecosystem through vignettes highlighting several BICCN analysis and visualization tools. Finally, we present emerging standards that have been developed or adopted by the BICCN toward FAIR (Wilkinson et al. 2016a) neuroscience. The combined BICCN ecosystem provides a comprehensive resource for the exploration and analysis of cell types in the brain.

Replication of the genome must be coordinated with gene transcription and cellular metabolism, especially following replication stress in the presence of limiting deoxyribonucleotides. TheS. cerevisiaeRad53 (CHEK2 in mammals) checkpoint kinase plays a major role in cellular responses to DNA replication stress. Cell cycle regulated, genome-wide binding of Rad53 to chromatin was examined. Under replication stress, the kinase bound to sites of active DNA replication initiation and fork progression, but unexpectedly to the promoters of about 20% of genes encoding proteins involved in multiple cellular functions. Rad53 promoter binding correlated with changes in expression of a subset of genes. Rad53 promoter binding to certain genes was influenced by sequence-specific transcription factors and less by checkpoint signaling. However, in checkpoint mutants, untimely activation of late-replicating origins reduces the transcription of nearby genes, with concomitant localization of Rad53 to their gene bodies. We suggest that the Rad53 checkpoint kinase coordinates genome-wide replication and transcription under replication stress conditions.

Enhanced cognitive function in humans is hypothesized to result from cortical expansion and increased cellular diversity. However, the mechanisms that drive these phenotypic differences remain poorly understood, in part due to the lack of high-quality cellular resolution data in human and non-human primates. Here, we take advantage of single cell expression data from the middle temporal gyrus of five primates (human, chimp, gorilla, macaque and marmoset) to identify 57 homologous cell types and generate cell-type specific gene coexpression networks for comparative analysis. While ortholog expression patterns are generally well conserved, we find 24% of genes with extensive differences between human and non-human primates (3383/14,131), which are also associated with multiple brain disorders. To validate these observations, we perform a meta-analysis of coexpression networks across 19 animals, and find that a subset of these genes have deeply conserved coexpression across all non-human animals, and strongly divergent coexpression relationships in humans (139/3383, NHEJ1, GTF2H2, C2 and BBS5) typically evolve under relaxed selective constraints and may drive rapid evolutionary change in brain function.One Sentence SummaryCross-primate middle temporal gyrus single cell expression data reveals patterns of conservation and divergence that can be validated with population coexpression networks.

Humans have unique cognitive abilities among primates, including language, but their molecular, cellular, and circuit substrates are poorly understood. We used comparative single nucleus transcriptomics in adult humans, chimpanzees, gorillas, rhesus macaques, and common marmosets from the middle temporal gyrus (MTG) to understand human-specific features of cellular and molecular organization. Human, chimpanzee, and gorilla MTG showed highly similar cell type composition and laminar organization, and a large shift in proportions of deep layer intratelencephalic-projecting neurons compared to macaque and marmoset. Species differences in gene expression generally mirrored evolutionary distance and were seen in all cell types, although chimpanzees were more similar to gorillas than humans, consistent with faster divergence along the human lineage. Microglia, astrocytes, and oligodendrocytes showed accelerated gene expression changes compared to neurons or oligodendrocyte precursor cells, indicating either relaxed evolutionary constraints or positive selection in these cell types. Only a few hundred genes showed human-specific patterning in all or specific cell types, and were significantly enriched near human accelerated regions (HARs) and conserved deletions (hCONDELS) and in cell adhesion and intercellular signaling pathways. These results suggest that relatively few cellular and molecular changes uniquely define adult human cortical structure, particularly by affecting circuit connectivity and glial cell function.

Single cell RNA-sequencing data have significantly advanced the characterization of cell type diversity and composition. However, cell type definitions vary across data and analysis pipelines, raising concerns about cell type validity and generalizability. With MetaNeighbor, we proposed an efficient and robust quantification of cell type replicability that preserves dataset independence and is highly scalable compared to dataset integration. In this protocol, we show how MetaNeighbor can be used to characterize cell type replicability by following a simple 3-step procedure: gene filtering, neighbor voting, and visualization. We show how these steps can be tailored to quantify cell type replicability, determine gene sets that contribute to cell type identity, and pretrain a model on a reference taxonomy to rapidly assess newly generated data. The protocol is based on an open-source R package available from Bioconductor and Github, requires basic familiarity with Rstudio or the R command line, and can typically be run in less than 5 minutes for millions of cells.

Different plant species within the grasses were parallel targets of domestication, giving rise to crops with distinct evolutionary histories and traits. Key traits that distinguish these species are mediated by specialized cell types within organs. Here, we compare the transcriptomes of all cells within roots in three grasses—Zea mays (maize), Sorghum bicolor (sorghum), and outgroup Setaria viridis (Setaria). We first show that single-cell and single-nucleus RNA-seq provide complementary readouts of cell identity, warranting a combined analysis. Comparative cellular analysis shows that the transcriptomes of some cell types diverged more rapidly than others, in part by recruiting gene modules from other cell types. Furthermore, examining the whole genome duplication in maize, we detect extensive dosage compensation in surviving co-expressed homeologs, reinforcing genomic balance1. Homeolog pairs that underwent subfunctionalization2, partitioning their expression among cell types, represented a minor pattern but showed the highest rate of acquiring a novel (non-ancestral) domain. These results fit a conjecture in which mechanisms that maintain stoichiometric balance at the molecular level aid in homeolog retention for extended periods to allow new functions to arise. An unexpected synergy between spatial sub- and neo-functionalization then contributes to changes in transcriptional cell identity.

Functional circuits consist of neurons with diverse axonal projections and gene expression. Understanding the molecular signature of projections requires high-throughput interrogation of both gene expression and projections to multiple targets in the same cells at cellular resolution, which is difficult to achieve using current technology. Here, we introduce BARseq2, a technique that simultaneously maps projections and detects multiplexed gene expression by in situ sequencing. We determined the expression of cadherins and cell-type markers in 29,933 cells, and the projections of 3,164 cells in both the mouse motor cortex and auditory cortex. Associating gene expression and projections in 1,349 neurons revealed shared cadherin signatures of homologous projections across the two cortical areas. These cadherins were enriched across multiple branches of the transcriptomic taxonomy. By correlating multi-gene expression and projections to many targets in single neurons with high throughput, BARseq2 provides a potential path to uncovering the molecular logic underlying neuronal circuits.

SummaryAlzheimer’s disease (AD), the leading cause of dementia, affects millions of people worldwide. With no disease-modifying medication currently available, the human toll and economic costs are rising rapidly. Under current standards, a patient is diagnosed with AD when both cognitive decline and pathology (amyloid plaques and neurofibrillary tangles) are present. Remarkably, some individuals who have AD pathology remain cognitively normal. Uncovering factors that lead to “cognitive resilience” to AD is a promising path to create new targets for therapies. However, technical challenges discovering novel human resilience factors limit testing, validation, and nomination of novel drugs for AD. In this study, we use single-nucleus transcriptional profiles of postmortem cortex from human individuals with high AD pathology who were either cognitively normal (resilient) or cognitively impaired (susceptible) at time of death, as well as mouse strains that parallel these differences in cognition with high amyloid load. Our cross-species discovery approach highlights a novel role for excitatory layer 4/5 cortical neurons in promoting cognitive resilience to AD, and nominates several resilience genes that includeATP1A1,GRIA3,KCNMA1, andSTXBP1. This putative cell type has been implicated in resilience in previous studies on bulk RNA-seq tissue, but our single-nucleus and cross-species approach identifies particular resilience-associated gene signatures in these cells. These novel resilience candidate genes were tested for replication in orthogonal data sets and confirmed to be correlated with cognitive resilience. Based on these gene signatures, we identified several potential mechanisms of resilience, including regulation of synaptic plasticity, axonal and dendritic development, and neurite vesicle transport along microtubules that are potentially targetable by available therapeutics. Because our discovery of resilience-associated genes in layer 4/5 cortical neurons originates from an integrated human and mouse transcriptomic space from susceptible and resilient individuals, we are positioned to test causality and perform mechanistic, validation, and pre-clinical studies in our human-relevant AD-BXD mouse panel.

What makes a mouse a mouse, and not a hamster? Differences in gene regulation between the two organisms play a critical role. Comparative analysis of gene coexpression networks provides a general framework for investigating the evolution of gene regulation across species. Here, we compare coexpression networks from 37 species and quantify the conservation of gene activity 1) as a function of evolutionary time, 2) across orthology prediction algorithms, and 3) with reference to cell- and tissue-specificity. We find that ancient genes are expressed in multiple cell types and have well conserved coexpression patterns, however they are expressed at different levels across cell types. Thus, differential regulation of ancient gene programs contributes to transcriptional cell identity. We propose that this differential regulation may play a role in cell diversification in both the animal and plant kingdoms.

Analysis of patient-derived DNA samples has identified hundreds of variants that are likely involved in neuropsychiatric diseases such as autism spectrum disorder (ASD) and schizophrenia (SCZ). While these studies couple behavioral phenotypes to individual genotypes, the number and diversity of candidate genes implicated in these disorders highlights the fact that the mechanistic underpinnings of these disorders are largely unknown. Here, we describe a RNAi-based screening platform that uses C. elegans to screen candidate neuropsychiatric risk genes (NRGs) for roles in controlling dendritic arborization. To benchmark this approach, we queried published lists of NRGs whose variants in ASD and SCZ are predicted to result in complete or partial loss of gene function. We found that a significant fraction (>16%) of these candidate NRGs are essential for dendritic development. Furthermore, these gene sets are enriched for dendritic arbor phenotypes (>14 fold) when compared to control RNAi datasets of over 500 human orthologs. The diversity of PVD structural abnormalities observed in these assays suggests that the functions of diverse NRGs (encoding transcription factors, chromatin remodelers, molecular chaperones and cytoskeleton-related proteins) converge to regulate neuronal morphology and that individual NRGs may play distinct roles in dendritic branching. We also demonstrate that the experimental value of this platform by providing additional insights into the molecular frameworks of candidate NRGs. Specifically, we show that ANK2/UNC-44 function is directly integrated with known regulators of dendritic arborization and suggest that altering the dosage of ARID1B/LET-526 expression during development affects neuronal morphology without diminishing aspects of cell fate specification.

Charting an organs’ biological atlas requires us to spatially resolve the entire single-cell transcriptome, and to relate such cellular features to the anatomical scale. Single-cell and single-nucleus RNA-seq (sc/snRNA-seq) can profile cells comprehensively, but lose spatial information. Spatial transcriptomics allows for spatial measurements, but at lower resolution and with limited sensitivity. Targeted in situ technologies solve both issues, but are limited in gene throughput. To overcome these limitations we present Tangram, a method that aligns sc/snRNA-seq data to various forms of spatial data collected from the same region, including MERFISH, STARmap, smFISH, Spatial Transcriptomics (Visium) and histological images. Tangram can map any type of sc/snRNA-seq data, including multimodal data such as those from SHARE-seq, which we used to reveal spatial patterns of chromatin accessibility. We demonstrate Tangram on healthy mouse brain tissue, by reconstructing a genome-wide anatomically integrated spatial map at single-cell resolution of the visual and somatomotor areas., Tangram is a versatile tool for aligning single-cell and single-nucleus RNA-seq data to spatially resolved transcriptomics data using deep learning.

X-chromosome inactivation (XCI) is a random, permanent, and developmentally early epigenetic event that occurs during mammalian embryogenesis. We harness these features to investigate characteristics of early lineage specification events during human development. We initially assess the consistency of X-inactivation and establish a robust set of XCI-escape genes. By analyzing variance in XCI ratios across tissues and individuals, we find that XCI is shared across all tissues, suggesting that XCI is completed in the epiblast (in at least 6-16 cells) prior to specification of the germ layers. Additionally, we exploit tissue-specific variability to characterize the number of cells present during tissue-lineage commitment, ranging from approximately 20 cells in liver and whole blood tissues to 80 cells in brain tissues. By investigating the variability of XCI ratios using adult tissue, we characterize embryonic features of human XCI and lineage specification that are otherwise difficult to ascertain experimentally.

The primary motor cortex (M1) is essential for voluntary fine-motor control and is functionally conserved across mammals1. Here, using high-throughput transcriptomic and epigenomic profiling of more than 450,000 single nuclei in humans, marmoset monkeys and mice, we demonstrate a broadly conserved cellular makeup of this region, with similarities that mirror evolutionary distance and are consistent between the transcriptome and epigenome. The core conserved molecular identities of neuronal and non-neuronal cell types allow us to generate a cross-species consensus classification of cell types, and to infer conserved properties of cell types across species. Despite the overall conservation, however, many species-dependent specializations are apparent, including differences in cell-type proportions, gene expression, DNA methylation and chromatin state. Few cell-type marker genes are conserved across species, revealing a short list of candidate genes and regulatory mechanisms that are responsible for conserved features of homologous cell types, such as the GABAergic chandelier cells. This consensus transcriptomic classification allows us to use patch–seq (a combination of whole-cell patch-clamp recordings, RNA sequencing and morphological characterization) to identify corticospinal Betz cells from layer 5 in non-human primates and humans, and to characterize their highly specialized physiology and anatomy. These findings highlight the robust molecular underpinnings of cell-type diversity in M1 across mammals, and point to the genes and regulatory pathways responsible for the functional identity of cell types and their species-specific adaptations., An examination of motor cortex in humans, marmosets and mice reveals a generally conserved cellular makeup that is likely to extend to many mammalian species, but also differences in gene expression, DNA methylation and chromatin state that lead to species-dependent specializations.

X-chromosome inactivation (XCI) is a random, permanent, and developmentally early epigenetic event that occurs during mammalian embryogenesis. We harness these features of XCI to investigate characteristics of early lineage specification events during human development. We initially assess the consistency of X-inactivation and establish a robust set of XCI-escape genes. By analyzing variance in XCI ratios across tissues and individuals, we find that XCI is completed prior to tissue specification and at a time when 6-16 cells are fated for all tissue lineages. Additionally, we exploit tissue specific variability to characterize the number of cells present at the time of each tissue’s lineage commitment, ranging from approximately 20 cells in liver and whole blood tissues to 80 cells in brain tissues. By investigating variance of XCI ratios using adult tissue, we resolve key features of human development otherwise difficult to ascertain experimentally and develop scalable methods easily applicable to future data.

Machine learning in genomics plays a key role in leveraging high-throughput data, but assessing the generalizability of performance has been a persistent challenge. Here, we propose to evaluate the generalizability of gene characterizations through the shape of performance curves. We identify Functional Equivalence Classes (FECs), uniform subsets of annotated and unannotated genes that jointly drive performance, by assessing the presence of straight lines in ROC curves. FECs are widespread across modalities and methods, and can be used to evaluate the extent and context-specificity of functional annotations in a data-driven manner. For example, FECs suggest that B cell markers can be decomposed into shared primary markers (10 to 50 genes), and tissue-specific secondary markers (100 to 500□genes). In addition, FECs are compatible with a wide range of functional encodings, with marker sets spanning at most 5% of the genome and data-driven extensions of Gene Ontology sets spanning up to 40% of the genome. Simple to assess visually and statistically, the identification of FECs in performance curves paves the way for novel functional characterization and increased robustness in analysis.

The clinical importance of the hematopoietic system makes it one of the most heavily studied lineages in all of biology. A clear understanding of the cell types and functional programs during hematopoietic development is central to research in aging, cancer, and infectious diseases. Known cell types are traditionally identified by the expression of proteins on the surface of the cells. Stem and progenitor cells defined based on these markers are assigned functions based on their lineage potential. The rapid growth of single cell RNA sequencing technologies (scRNAseq) provides a new modality for evaluating the cellular and functional landscape of hematopoietic stem and progenitor cells. The popularity of this technology among hematopoiesis researchers enables us to conduct a robust meta-analysis of mouse bone marrow scRNAseq data. Using over 300,000 cells across 12 datasets, we evaluate the classification and function of cell types based on discrete clustering,in silicoFACS sorting, and a continuous trajectory. We identify replicable signatures that define cell types based on genes and known cellular functions. Additionally, we evaluate the conservation of signatures associated with erythroid and monocyte lineage development across species using co-expression networks. The co-expression networks predict the effectiveness of the signature at identifying erythroid and monocyte cells in zebrafish and human scRNAseq data. Together, this analysis provides a robust reference, particularly marker genes and functional annotations, for future experiments in hematopoietic development.Key PointsMeta-analysis of 9 mouse bone marrow scRNAseq identifies markers for cell types and hematopoietic developmentCharacterize a replicable functional landscape of cell types by exploiting co-expression

Significance The identification of genes that are differentially expressed provides a molecular foothold onto biological questions of interest. Whether some genes are more likely to be differentially expressed than others, and to what degree, has never been assessed on a global scale. Here, we reanalyze more than 600 studies and find that knowledge of a gene’s prior probability of differential expression (DE) allows for accurate prediction of DE hit lists, regardless of the biological question. This result suggests redundancy in transcriptomics experiments that both informs gene set interpretation and highlights room for growth within the field., Differential expression (DE) is commonly used to explore molecular mechanisms of biological conditions. While many studies report significant results between their groups of interest, the degree to which results are specific to the question at hand is not generally assessed, potentially leading to inaccurate interpretation. This could be particularly problematic for metaanalysis where replicability across datasets is taken as strong evidence for the existence of a specific, biologically relevant signal, but which instead may arise from recurrence of generic processes. To address this, we developed an approach to predict DE based on an analysis of over 600 studies. A predictor based on empirical prior probability of DE performs very well at this task (mean area under the receiver operating characteristic curve, ∼0.8), indicating that a large fraction of DE hit lists are nonspecific. In contrast, predictors based on attributes such as gene function, mutation rates, or network features perform poorly. Genes associated with sex, the extracellular matrix, the immune system, and stress responses are prominent within the “DE prior.” In a series of control studies, we show that these patterns reflect shared biology rather than technical artifacts or ascertainment biases. Finally, we demonstrate the application of the DE prior to data interpretation in three use cases: (i) breast cancer subtyping, (ii) single-cell genomics of pancreatic islet cells, and (iii) metaanalysis of lung adenocarcinoma and renal transplant rejection transcriptomics. In all cases, we find hallmarks of generic DE, highlighting the need for nuanced interpretation of gene phenotypic associations.

Gene-gene relationships are commonly measured via the co-variation of gene expression across samples, also known as gene co-expression. Because shared expression patterns are thought to reflect shared function, co-expression networks describe functional relationships between genes, including co-regulation. However, the heterogeneity of cell types in bulk RNA-seq samples creates connections in co-expression networks that potentially obscure co-regulatory modules. The brain initiative cell census network (BICCN) single-cell RNA sequencing (scRNA-seq) datasets provide an unparalleled opportunity to understand how gene-gene relationships shape cell identity. Comparison of the BICCN data (500,000 cells/nuclei across 7 BICCN datasets) with that of bulk RNA-seq networks (2,000 mouse brain samples across 52 studies) reveals a consistent topology reflecting a shared co-regulatory signal. Differential signals between broad cell classes persist in driving variation at finer levels, indicating that convergent regulatory processes affect cell phenotype at multiple scales.