CoChat
Sign Up
Cited by 663Functional Brain Connectivity StudiesSingle-cell and spatial transcriptomicsBioinformatics and Genomic Networks

Canonical genetic signatures of the adult human brain

Open access
Hawrylycz, Michael, Miller, Jeremy A, Menon, Vilas, Feng, David, Dolbeare, Tim, Guillozet-Bongaarts, Angela L, Jegga, Anil G, Aronow, Bruce J +22 more·2015·Nature Neuroscience
DOI 10.1038/nn.4171PMC PMC4700510
Open this paper in CoChat — free account

Abstract

The structure and function of the human brain are highly stereotyped, implying a conserved molecular program responsible for its development, cellular structure and function. We applied a correlation-based metric called differential stability to assess reproducibility of gene expression patterning across 132 structures in six individual brains, revealing mesoscale genetic organization. The genes with the highest differential stability are highly biologically relevant, with enrichment for brain-related annotations, disease associations, drug targets and literature citations. Using genes with high differential stability, we identified 32 anatomically diverse and reproducible gene expression signatures, which represent distinct cell types, intracellular components and/or associations with neurodevelopmental and neurodegenerative disorders. Genes in neuron-associated compared to non-neuronal networks showed higher preservation between human and mouse; however, many diversely patterned genes displayed marked shifts in regulation between species. Finally, highly consistent transcriptional architecture in neocortex is correlated with resting state functional connectivity, suggesting a link between conserved gene expression and functionally relevant circuitry.

AI snapshot

Written from the paper’s full text.

TL;DR

The paper applies a correlation-based “differential stability” (DS) metric to Allen Human Brain Atlas expression data across 132 structures in six adult human brains to identify conserved transcriptional organization. It reports 32 reproducible co-expression modules, links high-DS genes to brain function, disease, and drug targets, compares module conservation with mouse, and finds that conserved cortical gene expression correlates with resting-state functional connectivity.

Key findings

  • DS ranged from ΔBR=0.970 to −0.057 across 17,348 uniquely annotated human transcripts; the highest-DS gene was FOXG1 (ΔBR=0.970), and PCDH8 had ΔBR=0.931. The top 10% DS genes (n=1735) were enriched for neuronal functional categories, drug interactions (first decile had 4651 small-molecule/chemical interactions, more than 30 times the third decile), and disease gene sets including mental disorders (p<8.06e-31), schizophrenia (p<9.14e-30), autism (p<3.45e-17), epilepsy (p<1.00e-11), bipolar disorder (p<4.94e-14), and major depression (p<4.60e-10).
  • A consensus WGCNA network built from the 50% highest DS genes (ΔBR>0.5284, g=8674) identified 32 modules; 90.1% (15,627) of genes correlated with these patterns at kME>0.4. Modules ranged from largely neuronal (M1–M15) to essentially non-neuronal (M16–M32), with examples including M6 in cortex, M10 in striatum, M11 in thalamus, M12 in substantia nigra, M14 in hypothalamus, M16 in dentate gyrus, and M17 in cerebellum; M25 was associated with immune response (q<3.71e-35) and M30 with myelination/glial ensheathment (q<1.48e-06).
  • In human-mouse comparison using 53 matched mouse structures and 2651 genes, neuron-associated modules such as M1, M6, and M7 were better preserved than many non-neuronal modules (M25–M32), though individual genes within even preserved modules sometimes shifted patterning between species.
  • In cortex, the top 5th percentile of cortical DS genes (n=867) had ΔCTX>0.357 versus ΔBR>0.854 for whole brain, and only 31% of the top 5% cortical DS genes were in the top 5% of brain-wide DS. Higher cortical DS predicted stronger correlation between functional connectivity and gene expression (ρ=0.46, p<2.2e-16); 136 genes whose polymorphisms affect resting-state functional connectivity had higher expression-connectivity correlation (μ=0.12, p<2.2e-192) and higher cortical DS (ΔCTX=0.40, p<1.2e-325).

Methods

  • Data came from the Allen Human Brain Atlas: six neurotypical adult whole brains (3 Caucasian males, 2 African American males, 1 Caucasian woman), with 345–911 samples per brain spanning one or both hemispheres, 232 discrete brain structures sampled at least once, and 132 structures used for the main analysis; expression was profiled with whole-genome Agilent microarrays.
  • DS was defined as the average Pearson correlation across 15 pairs of the six brains over a set of anatomic regions; WGCNA was used to construct a consensus co-expression network from the top 50% DS genes, with modules assigned by kME>0.4 and functionally annotated using ToppGene and related databases.
  • Cross-species comparison used Allen Mouse Brain Atlas in situ hybridization expression energy for 2651 genes across 53 matched mouse structures; functional connectivity analysis used Human Connectome Project resting-state fMRI data, including a dense connectome from 468 subjects and parcellated connectomes from 447 subjects across 52 left-hemisphere parcels, then correlated gene co-expression matrices with the connectome matrix.

Limitations

  • The paper states possible ascertainment bias in regional transcriptional patterns because they may be preferentially annotated in available studies, and notes that approximately 9% of the most conserved gene set has no functional annotation to date.
  • The paper states that the mouse comparison used 2651 genes initially chosen because they were potentially biologically interesting, making that set biased; specific proportions of cross-species agreement may not represent the whole transcriptome, though gene assignments as conserved or not should reflect actual conservation in these datasets.
  • The paper states that functional connectivity from fMRI is an indirect measure and should not be considered a surrogate for direct anatomical connectivity; it also notes that anatomical sampling was not cell type-specific, and that direct comparisons with previous studies are challenging due to differences in analysis strategies and experimental design.

Model-written summary. Check it against the paper before relying on it.

Integrity signals

Registry

Crossref · 10.1038/nn.4171

Cites

49 of 49 references verified0 citation errors · 0 not checkable

Statistics

Nothing to recompute.

Claims the paper makes

0 of 21 cited sources checked · 0 supported · 0 flagged

  • Findingown data

    Genes in neuron-associated compared to non-neuronal networks showed higher preservation between human and mouse.

    “Genes in neuron-associated compared to non-neuronal networks showed higher preservation between human and mouse; however, many diversely-patterned genes displayed dramatic shifts in regulation between species.”
    Quote in the paperSubstantial claim
  • Findingown data

    Applying a correlation-based metric of differential stability (DS) to gene expression across 132 structures in six individual brains reveals meso-scale genetic organization.

    “We applied a correlation-based metric of “differential stability” (DS) to assess reproducibility of gene expression patterning across 132 structures in six individual brains, revealing meso-scale genetic organization.”
    Quote in the paperSubstantial claim
  • Findingown data

    Using high DS genes, the authors identified 32 anatomically diverse and reproducible gene expression signatures.

    “Using high DS genes we identified 32 anatomically diverse and reproducible gene expression signatures, which represent distinct cell types, intracellular components, and/or associations with neurodevelopmental and neurodegenerative disorders.”
    Quote in the paperSubstantial claim
  • Findingown data

    Large transcriptional differences are seen between major brain structures.

    “Large transcriptional differences are seen between major structures.”
    Quote in the paperSubstantial claim
  • Findingown data

    Using the six brains, the authors identified a consensus network common to all brains using the 50% highest DS genes (ΔBR>0.5284, g=8674).

    “Using the six brains now available, we identified a consensus network22 common to all brains (Fig. 3a, panel 1), using the 50% highest DS genes (ΔBR>0.5284, g=8674, Supplementary Methods).”
    Quote in the paperSubstantial claim0 of 1 sources checked
    • cites [22] Langfelder, P., Luo, R., Oldham, M. & Horvath, S. Is my network module preserved and reproducible? PLoS Comput. Biol. 7, e1001057 (2011). source

      Not checked in the free run.

  • May combine two claims. 35
  • The quote only partly states this claim. 3
  • Classified as probably not a substantial claim. 22

  • May combine two claims.

    Findingown data

    The highest DS genes are enriched for brain-related biological annotations, disease associations, drug targets, and literature citations.

    “The highest DS genes are highly biologically relevant, with enrichment for brain-related biological annotations, disease associations, drug targets, and literature citations.”
    Quote in the paperSubstantial claim

  • May combine two claims.

    Findingown data

    Potassium channels are the most over-represented among high DS genes (p<1.70e-12), exhibiting diverse and specific expression patterns.

    “Interestingly, potassium channels (p<1.70e-12) are the most over-represented among high DS genes, exhibiting diverse and specific expression patterns (Supplementary Fig. 1).”
    Quote in the paperSubstantial claim

  • May combine two claims.

    Findingown data

    A smaller percentage of genes have high DS in cortex than in whole brain; the top 5th percentile (n=867) of DS cortex genes had ΔCTX>0.357 versus ΔBR>0.854 for whole brain, and the rank ordering of genes by whole-brain DS vs cortex-only DS differs markedly, with only 31% of the top 5% of cortical DS genes in the top 5% of brain-wide DS.

    “A smaller percentage of genes have high DS in cortex (Fig. 7b); in contrast to whole brain, the top 5th percentile (n=867) of DS cortex genes had ΔCTX>0.357 versus ΔBR>0.854 for whole brain. The rank ordering of genes by whole-brain DS vs cortex-only DS differs markedly, with only 31% of the top 5% of cortical DS genes in the top 5% of brain-wide DS, (Fig. 7b).”
    Quote in the paperSubstantial claim

  • May combine two claims.

    Findingown data

    136 genes whose polymorphisms significantly affect resting-state functional connectivity in a large sample of healthy adolescents have both higher correlation of expression with functional connectivity (μ=0.12, p<2.2e-192) and higher DS (ΔCTX=0.40, p<1.2e-325) in human cortex, though they constitute only a minority (15%) of the highest cortical 5th percentile DS genes.

    “Further, we found that 136 genes whose polymorphisms significantly affect resting-state functional connectivity in a large sample of healthy adolescents28 have both higher correlation of expression with functional connectivity (μ=0.12, p<2.2e-192) and higher DS (ΔCTX=0.40, p<1.2e-325) in human cortex (red points, Fig. 7d), though they constitute only a minority (15%) of the highest cortical 5th p”
    Quote in the paperSubstantial claim0 of 1 sources checked
    • cites [28] Richiardi, J. et al. Correlated gene expression supports synchronous activity in brain networks. Science 348, 1241–1244 (2015). source

      Not checked in the free run.

  • May combine two claims. The quote only partly states this claim.

    Findingown data

    A small number (302) of high DS genes appear to have unique spatial expression patterns and were not strongly correlated to any of the 32 module eigengenes; among these are extensively studied genes and regional markers; these genes are enriched in GPCR pathway and signaling events and over-represented for SNP-associated neurotransmission and neurodevelopment genes conferring susceptibility to某些 等

    “Finally, a small number (302) of high DS genes appear to have unique spatial expression patterns, and were not strongly correlated to any of the 32 module eigengenes. 50 of these genes are shown in Figure 5, illustrating the diversity of their expression patterns.”
    Quote in the paperSubstantial claim

  • May combine two claims.

    Findingown data

    From a global anatomical perspective, the 32 gene modules were predominantly enriched either in the telencephalon or in non-telencephalic deeper brain structures, and individual modules were enriched in distinct combinations of structures and associated with distinct functional pathways, disease and drug interactions.

    “From a global anatomical perspective, the 32 gene modules were predominantly enriched either in the telencephalon or in non-telencephalic deeper brain structures (Fig. 4a). Individual modules were enriched in distinct combinations of structures (Fig. 4b; Supplementary Figs. 5–8) and were associated with distinct functional pathways, disease and drug interactions (Table 1; Supplementary Fig. 9).”
    Quote in the paperSubstantial claim

  • May combine two claims.

    Findingown data

    Low DS genes with ubiquitous expression are typically not brain specific and are enriched for housekeeping functions such as RNA binding, KEGG spliceosome pathway, and mitochondrial ribosomal proteins.

    “In contrast, low DS genes with ubiquitous expression (lower 25th percentile in variability across brain regions, below the horizontal green line in Fig. 2b; Supplementary Table 3), such as the ubiquitin isoforms UBB and UBC, are typically not brain specific. Instead, they appear to provide the basal machinery common to all cell and are enriched for housekeeping functions such as RNA binding (p<3.”
    Quote in the paperSubstantial claim

  • May combine two claims.

    Findingown data

    There was a clear ordering of modules based on predominance of neuronal gene expression, and the proportions ranged widely from largely neuronal signatures (M1-M15) to essentially non-neuronal (M16-M32).

    “While not a perfect comparison given the differences between species and in vivo versus in vitro, there was a clear ordering of modules that could be seen based on predominance of neuronal gene expression (Fig. 3c). These proportions ranged widely from those with largely neuronal signatures (M1-M15) to essentially non-neuronal (M16-M32).”
    Quote in the paperSubstantial claim

  • May combine two claims.

    Findingown data

    Some modules were well preserved between human and mouse, including the neuronal-enriched modules M1, M6 and M7, whereas many of the most non-neuronal modules (M25-M32) were poorly preserved.

    “Some modules were well preserved (Fig. 6a), including the neuronal-enriched modules M1, M6 and M7, whereas many of the most non-neuronal modules (M25-M32) were poorly preserved.”
    Quote in the paperSubstantial claim

  • May combine two claims.

    Findingown data

    Higher cortical DS of a gene is predictive of a stronger correlation between functional connectivity and cortical gene expression pattern (ρ = 0.46, p<2.2e-16), while the correlation with brain-wide DS alone was much weaker (ρ=0.17).

    “Fig. 7d shows that higher cortical DS of a gene is predictive of a stronger correlation between functional connectivity and cortical gene expression pattern (ρ = 0.46, p<2.2e-16; the correlation with brain-wide DS alone was much weaker (ρ=0.17), indicating that the conserved cortical genes correlate most strongly with cortical functional connectivity.”
    Quote in the paperSubstantial claim

  • May combine two claims.

    Findingown data

    Gene sets associated with mental disorders are significantly enriched in the top 10% of DS genes, in particular those for schizophrenia, autism, epilepsy, bipolar disorder, and major depression.

    “Genes sets associated with mental disorders (p<8.06e-31) are significantly enriched in the top 10% of DS genes (Fig. 2f; Supplementary Table 5), in particular those for schizophrenia (p<9.14e-30), autism (p<3.45e-17), epilepsy (p<1.00e-11), bipolar disorder (p<4.94e-14), and major depression (p<4.60e-10).”
    Quote in the paperSubstantial claim

  • May combine two claims. The quote only partly states this claim.

    Findingown data

    M12 is uniquely associated with dopamine biosynthesis and cocaine addiction; M19 is associated with mitochondria and demyelination and has strong disease associations with Alzheimer's, Parkinson's, and Leigh's syndrome; and M25 is enriched in the glia-rich corpus callosum and internal portion of the globus pallidus and is associated with immune response and systemic lupus erythemetosis.

    “In agreement with known function, M12 is uniquely associated with dopamine biosynthesis (q<8.39e-06) and cocaine addiction (q<5.64e-05). Other modules showed more complex anatomical patterning, such as non-telencephalic expression (M19), the inverse of M1. M19 is associated with mitochondria (q<5.50e-82) and demyelination (q<2.57e-04), and has strong disease associations with Alzheimer’s (q<3.98e-”
    Quote in the paperSubstantial claim

  • May combine two claims.

    Findingown data

    Individual genes with matching anatomical and cellular patterns could be found for each module, but for each module there were also genes whose pattern fundamentally shifts between species; even in highly preserved modules, some individual genes differ in their patterning across species, and the frequency of these changes varies by module with better general pattern conservation in more strongly n

    “Despite this variation in the overall strength of correlation across species, individual genes with matching anatomical and cellular patterns could be found for each module (left 2 ISH panels in Fig. 6c–g). On the other hand, for each module there were genes whose pattern fundamentally shifts between species (far right ISH panels), for example from thalamic to cerebellar in M11 or from glial to”
    Quote in the paperSubstantial claim

  • May combine two claims.

    Findingown data

    The great majority of all genes (90.1%, 15,627) are correlated with this relatively small set of patterns, and the MEs derived from these genes are highly stable.

    “The great majority of all genes (90.1%, 15,627) are correlated with this relatively small set of patterns (Supplementary Table 6), and the MEs derived from these genes are highly stable (Fig. 3b).”
    Quote in the paperSubstantial claim

  • May combine two claims. The quote only partly states this claim.

    Findingown data

    Over 96% of all annotations were associated with just 17 of the 32 modules; modules M1-M6 contain by far the most neuronal function-related annotations; and the most non-neuronal modules (M21-M32) show strong and unique associations with vasculature development, immune system function, and myelination and glial ensheathment.

    “Interestingly, over 96% of all annotations were associated with just 17 of the 32 modules (left-most in Fig. 3e) including those restricted to the highly studied nigro-striatal pathway structures associated with Parkinson’s disease (M10 and M12) as well as a set of predominantly non-neuronal modules (M24, M23, M25, M30; Table 1).”
    Quote in the paperSubstantial claim

  • May combine two claims.

    Findingown data

    The top DS genes have a disproportionate number of drug interactions and a monotonically decreasing association with miRNA targets.

    “Furthermore, the top DS genes have a disproportionate number of drug interactions and a monotonically decreasing association with miRNA targets.”
    Quote in the paperNot a substantial claim

  • May combine two claims.

    Findingown data

    Despite massive differences in size and measurement techniques of the two data modalities, many of the modules had very highly correlated expression across specific anatomical regions; for example, M6, M10, M11 and M17 had nearly identical patterning, whereas M30 was very poorly correlated.

    “Despite massive differences in size and measurement techniques of these two data modalities, many of the modules had very highly correlated expression across specific anatomical regions (Fig. 6c–g). For example, the forebrain-enriched and neuron-associated M6, striatal M10, thalamic M11 and cerebellar M17 had nearly identical patterning. In contrast, the module eigengenes for the non-neuronal mod”
    Quote in the paperSubstantial claim

  • Probably not a substantial claim: may not pass S3 (load-bearing).

    Findingnone given

    Approximately 9% of the most conserved gene set has no functional annotation to date.

    “Interestingly, approximately 9% of this most conserved gene set has no functional annotation to date, thus representing a set of rich targets for future research into human brain function.”
    Quote in the paperNot a substantial claim

  • May combine two claims.

    Findingown data

    The highest 10% of brain-wide DS genes are substantially overrepresented for known drug targets and are highly significantly enriched for functional annotations of many cellular functions and biological pathways.

    “Remarkably, the highest 10% of brain-wide DS genes are substantially overrepresented for known drug targets and are highly significantly enriched for functional annotations of many cellular functions and biological pathways.”
    Quote in the paperSubstantial claim

  • May combine two claims.

    Findingown data

    The final set of networks showed consistent but more complex distributed anatomical patterning and were enriched in genes associated with organelles or cellular machinery such as mitochondria, the proteasome, ribonucleoprotein complexes and chromatin.

    “The final set of networks were more difficult to interpret, showing consistent but more complex distributed anatomical patterning and enriched in genes associated with organelles or cellular machinery such as mitochondria, the proteasome, ribonucleoprotein complexes and chromatin.”
    Quote in the paperSubstantial claim

  • May combine two claims.

    Findingown data

    Significant ontological categories of the highest cortical 5th percentile DS genes include main axon, neuron projection, calcium and sodium channel activity, and axon guidance Reactome pathway.

    “Significant ontological categories of this highest cortical 5th percentile DS genes include main axon (p<7.84e-09), neuron projection (p<2.2e-04), calcium and sodium channel activity (p<3.46e-11), and axon guidance Reactome pathway (p<2.22e-04).”
    Quote in the paperNot a substantial claim

  • May combine two claims.

    Findingnone given

    Many of those genes are well studied as markers of specific cell types (e.g. VIP and SST for GABAergic interneurons) or developmental zones (homeobox genes) and are not representative of the patterning of the great majority of genes.

    “Surprisingly, many of those genes are well studied as markers of specific cell types (e.g. VIP and SST for GABAergic interneurons) or developmental zones (homeobox genes) and are not representative of the patterning of the great majority of genes.”
    Quote in the paperSubstantial claim

  • May combine two claims.

    Interpretationown data

    The 32 gene expression signatures represent distinct cell types, intracellular components, and/or associations with neurodevelopmental and neurodegenerative disorders.

    “Using high DS genes we identified 32 anatomically diverse and reproducible gene expression signatures, which represent distinct cell types, intracellular components, and/or associations with neurodevelopmental and neurodegenerative disorders.”
    Quote in the paperSubstantial claim

  • Classified as probably not a substantial claim.

    Premisecitation

    Functional brain networks determined using resting-state fMRI can be recapitulated using correlated gene expression in postmortem brain tissue.

    “Recent work has revealed that functional brain networks determined using resting-state functional magnetic resonance imaging (fMRI) can be recapitulated using correlated gene expression in postmortem brain tissue28.”
    Quote in the paperSubstantial claim0 of 1 sources checked
    • cites [28] Richiardi, J. et al. Correlated gene expression supports synchronous activity in brain networks. Science 348, 1241–1244 (2015). source

      Not checked in the free run.

  • May combine two claims.

    Interpretationown data

    These patterns are likely to be representative of a larger population as they are highly consistent across all six individuals.

    “Furthermore, these patterns are likely to be representative of a larger population as they are highly consistent across all six individuals, and patterns identified using any five brains could be found reproducibly in the sixth (Supplementary Methods; Supplementary Fig. 3).”
    Quote in the paperSubstantial claim

  • May combine two claims.

    Premisecitation

    Human neocortex shows variations in transcriptional patterns that are consistent across individuals and are correlated with topographical position across the cortical sheet.

    “Despite its relative homogeneity compared to other brain regions, human neocortex shows variations in transcriptional patterns that are consistent across individuals and are correlated with topographical position across the cortical sheet3.”
    Quote in the paperSubstantial claim0 of 1 sources checked
    • cites [3] Hawrylycz, M.J. et al. An anatomically comprehensive atlas of the adult human brain transcriptome. Nature 489, 391–399 (2012). source

      Not checked in the free run.

  • Probably not a substantial claim: may not pass S3 (load-bearing).

    Interpretationreasoning

    This shift may represent a mechanism for species differentiation where mutations lead to dramatic shifts in gene regulation across brain regions.

    “This may represent a mechanism for species differentiation where mutations lead to dramatic shifts in gene regulation across brain regions.”
    Quote in the paperNot a substantial claim

  • May combine two claims.

    Interpretationreasoning

    These networks may represent core intracellular pathways that tend to be co-expressed and also differentially enriched in particular brain regions.

    “These networks may represent core intracellular pathways that tend to be co-expressed and also differentially enriched in particular brain regions.”
    Quote in the paperNot a substantial claim

  • Probably not a substantial claim: may not pass S3 (load-bearing).

    Interpretationreasoning

    Differences in analysis strategies and experimental design make direct comparisons with previous studies challenging.

    “Despite similarities in data type and methodology, these differences in analysis strategies and experimental design make direct comparisons with previous studies challenging.”
    Quote in the paperNot a substantial claim

  • Probably not a substantial claim: may not pass S3 (load-bearing).

    Premisecitation

    A study of ~700 genes in human versus mouse cortex by in situ hybridization found approximately 25% of genes having at least subtle differences in areal or cellular patterning.

    “a study of ~700 genes in human versus mouse cortex by in situ hybridization found approximately 25% of genes having at least subtle differences in areal or cellular patterning39”
    Quote in the paperNot a substantial claim0 of 1 sources checked
    • cites [39] Zeng, H. et al. Large-scale cellular-resolution gene profiling in human neocortex reveals species-specific molecular signatures. Cell 149, 483–496 (2012). source

      Not checked in the free run.

  • Probably not a substantial claim: may not pass S3 (load-bearing). May combine two claims.

    Premisecitation

    Functional connectivity as measured using fMRI is a powerful but indirect measure of network organization that can provide high spatial resolution but should not be considered a surrogate for direct anatomical connectivity.

    “Functional connectivity as measured using fMRI is a powerful but indirect measure of network organization that can provide high spatial resolution but should not be considered a surrogate for direct anatomical connectivity29, 30.”
    Quote in the paperNot a substantial claim0 of 2 sources checked
    • cites [29] Van Essen, D.C. et al. The WU-Minn Human Connectome Project: an overview. Neuroimage 80, 62–79 (2013). source

      Not checked in the free run.

    • cites [30] Smith, S.M. et al. Functional connectomics from resting-state fMRI. Trends Cogn. Sci. 17, 666–682 (2013). source

      Not checked in the free run.

  • Probably not a substantial claim: may not pass S3 (load-bearing).

    Premisecitation

    Major differences in glial structure and function between human and mice have been described.

    “major differences in glial structure and function between human and mice have been described38”
    Quote in the paperNot a substantial claim0 of 1 sources checked
    • cites [38] Oberheim, N.A., Wang, X., Goldman, S. & Nedergaard, M. Astrocytic complexity distinguishes the human brain. Trends Neurosci. 29, 547–553 (2006). source

      Not checked in the free run.

  • Probably not a substantial claim: may not pass S3 (load-bearing).

    Premisecitation

    Topological overlap has been shown to be a more biologically relevant metric than adjacency in many cases.

    “Topological overlap has been shown to be a more biologically relevant metric than adjacency in many cases10.”
    Quote in the paperNot a substantial claim0 of 1 sources checked
    • cites [10] Zhang, B. & Horvath, S. A general framework for weighted gene co-expression network analysis. Stat. Appl. Genet. Mol. Biol. 4, 17 (2005). source

      Not checked in the free run.

  • May combine two claims.

    Interpretationown data

    These genes represent a functionally critical set whose transcriptional regulation is tightly controlled.

    “This suggests that these genes represent a functionally critical set whose transcriptional regulation is tightly controlled.”
    Quote in the paperSubstantial claim

  • Classified as probably not a substantial claim.

    Study design and conductnone given

    Within each batch, array-specific biases characterized by probe GC content, location in the chip, and experiment-wise mean intensity are corrected.

    “Within each batch, array-specific biases characterized by probe GC content, location in the chip, and experiment-wise mean intensity are corrected.”
    Quote in the paperSubstantial claim

  • May combine two claims.

    Interpretationreasoning

    A second set of networks show striking enrichment in specific brain regions such as the striatum, thalamus, cerebellum and substantia nigra, presumably reflecting unique and coherent expression signatures responsible for the functions of specific neuronal types in these regions.

    “A second set of networks show striking enrichment in specific brain regions such as the striatum, thalamus, cerebellum and substantia nigra, presumably reflecting unique and coherent expression signatures responsible for the functions of specific neuronal types in these regions.”
    Quote in the paperSubstantial claim

  • Classified as probably not a substantial claim.

    Study design and conductnone given

    Adjustment is made due to the bias introduced in sample quality and expression level by different dissection methods.

    “Adjustment is made due to the bias introduced in sample quality and expression level by different dissection methods.”
    Quote in the paperSubstantial claim

  • Classified as probably not a substantial claim. May combine two claims.

    Study design and conductcitation

    ToppGene was used to calculate p-values from hypergeometric tests, corrected for multiple comparisons.

    “ToppGene (https://toppgene.cchmc.org) portal24 was used to access an extensive list of databases, and to calculate p-values from hypergeometric tests, corrected for multiple comparisons.”
    Quote in the paperSubstantial claim0 of 1 sources checked
    • cites [24] Chen, J., Bardes, E.E., Aronow, B.J. & Jegga, A.G. ToppGene Suite for gene list enrichment analysis and candidate gene prioritization. Nucleic Acids Res. 37, W305–W311 (2009). source

      Not checked in the free run.

  • Classified as probably not a substantial claim.

    Study design and conductreasoning

    The same control samples are included in all batches to correct for non-biological systematic biases among batches.

    “The same control samples are included in all batches to serve this purpose.”
    Quote in the paperSubstantial claim

  • Probably not a substantial claim: may not pass S3 (load-bearing).

    Study design and conductnone given

    Within a brain, cross-batch correction is done so that the mean expression levels of control samples in the batches may be aligned.

    “Within a brain, cross-batch correction is done so that the mean expression levels of control samples in the batches may be aligned.”
    Quote in the paperNot a substantial claim

  • Probably not a substantial claim: may not pass S3 (load-bearing).

    Premisecitation

    The frequency of differences in gene expression between species appears to vary as a function of evolutionary distance.

    “the frequency of differences appears to vary as a function of evolutionary distance37”
    Quote in the paperNot a substantial claim0 of 1 sources checked
    • cites [37] Morris, J.A. et al. Divergent and nonuniform gene expression patterns in mouse brain. Proc. Natl. Acad. Sci. USA 107, 19049–19054 (2010). source

      Not checked in the free run.

  • May combine two claims.

    Study design and conductown data

    This study builds upon and complements initial results by sampling brain regions at relatively equal depth by decreasing representation of cortex and cerebellum in the analysis, and by performing a consensus network analysis that exclusively identifies common patterns across brains.

    “This study builds upon and complements our initial results in two important ways: first, we sample brain regions at relatively equal depth by decreasing representation of cortex and cerebellum in the analysis, and second, we perform a consensus network analysis, which exclusively identifies common patterns across brains.”
    Quote in the paperSubstantial claim

  • Probably not a substantial claim: may not pass S3 (load-bearing).

    Premisecitation

    A meta-analysis of microarray studies showed better preservation of neuronal gene co-expression compared to glial and disease-related genes.

    “a meta-analysis of microarray studies which showed better preservation of neuronal gene co-expression compared to glial and disease-related genes36”
    Quote in the paperNot a substantial claim0 of 1 sources checked
    • cites [36] Miller, J.A., Horvath, S. & Geschwind, D.H. Divergence of human and mouse brain transcriptome highlights Alzheimer disease pathways. Proc. Natl. Acad. Sci. USA 107, 12698–12703 (2010). source

      Not checked in the free run.

  • Probably not a substantial claim: may not pass S3 (load-bearing). May combine two claims.

    Study design and conductreasoning

    Across brains, brain-wise mean expressions as well as control samples' mean expression are aligned.

    “Across brains, brain-wise mean expressions as well as control samples’ mean expression are aligned.”
    Quote in the paperNot a substantial claim

  • Probably not a substantial claim: may not pass S3 (load-bearing). May combine two claims.

    Study design and conductnone given

    The method for constructing consensus networks is entirely automated, resulting in an unbiased collection of distinct, yet well-defined modules.

    “This method is entirely automated, resulting in an unbiased collection of distinct, yet well-defined modules.”
    Quote in the paperNot a substantial claim

  • May combine two claims.

    Study design and conductreasoning

    These genes were distributed across all of the human modules and therefore represent a fairly complete set of expression patterns at the whole-brain level.

    “these genes were distributed across all of the human modules and therefore represent a fairly complete set of expression patterns at the whole-brain level.”
    Quote in the paperNot a substantial claim
    • Rests on an assumption: The mouse ISH expression energy values are comparable to human microarray intensities for between-species correlations and network statistics.
    • Rests on an assumption: Genes in mouse can be assigned to the same modules as their human orthologs.

  • Classified as probably not a substantial claim.

    Study design and conductown data

    A significance value for each gene-connectivity comparison was obtained using randomized gene co-expression matrices.

    “We also obtained a significance value for each gene-connectivity comparison using the randomized gene co-expression matrices.”
    Quote in the paperSubstantial claim

  • Classified as probably not a substantial claim. May combine two claims.

    Study design and conductreasoning

    While the specific proportions of genes that agree between species may not be representative of the transcriptome as a whole, the assignment of genes as either conserved or not conserved should accurately reflect their actual conservation based on these two data sets.

    “While the specific proportions of genes that agree between species may not be representative of the transcriptome as a whole, the assignment of genes as either conserved or not conserved in the text should accurately reflect their actual conservation based on these two data sets.”
    Quote in the paperSubstantial claim
    • Rests on an assumption: The mouse ISH expression energy values are comparable to human microarray intensities for between-species correlations and network statistics.
    • Rests on an assumption: Genes in mouse can be assigned to the same modules as their human orthologs.

  • Classified as probably not a substantial claim.

    Study design and conductcitation

    Each of four 15-min resting state runs for each subject was cleaned (denoised) using 24-parameter motion regression followed by ICA+FIX denoising.

    “Each of four 15-min resting state runs for each subject in the group was cleaned (denoised) using 24-parameter motion regression followed by ICA+FIX denoising30, 48.”
    Quote in the paperSubstantial claim0 of 2 sources checked
    • cites [30] Smith, S.M. et al. Functional connectomics from resting-state fMRI. Trends Cogn. Sci. 17, 666–682 (2013). source

      Not checked in the free run.

    • cites [48] Salimi-Khorshidi, G. et al. Automatic denoising of functional MRI data: combining independent component analysis and hierarchical fusion of classifiers. Neuroimage 90, 449–468 (2014). source

      Not checked in the free run.

  • Probably not a substantial claim: may not pass S3 (load-bearing), S4 (based).

    Study design and conductnone given

    Statistical tests were chosen such that the data meet assumptions of the specific statistical test.

    “To the best of our knowledge, statistical tests were chosen such that the data meet assumptions of the specific statistical test.”
    Quote in the paperNot a substantial claim

  • May combine two claims.

    Study design and conductreasoning

    The 2651 genes used for mouse comparison were initially chosen to be run in coronal section because they were potentially biologically interesting, and therefore represent a biased set of genes.

    “these 2651 genes were initially chosen to be run in coronal section because they were potentially biologically interesting in some way, and therefore represent a biased set of genes.”
    Quote in the paperNot a substantial claim
    • Rests on an assumption: The mouse ISH expression energy values are comparable to human microarray intensities for between-species correlations and network statistics.
    • Rests on an assumption: Genes in mouse can be assigned to the same modules as their human orthologs.

  • Probably not a substantial claim: may not pass S3 (load-bearing).

    Study design and conductnone given

    All p-values in the module enrichment analysis were Bonferroni corrected for multiple hypergeometric test comparisons.

    “All p-values in this analysis were Bonferroni corrected for multiple hypergeometric test comparisons.”
    Quote in the paperNot a substantial claim
52 of 52

  • The paper relies on this without support.

    The structure and function of the human brain are highly stereotyped.

    “The structure and function of the human brain are highly stereotyped, implying a conserved molecular program responsible for its development, cellular structure, and function.”

  • The paper relies on this without support.

    The general anatomical positioning of and connectivity between brain regions is highly stereotyped between individuals.

    “Despite a range of brain sizes across individuals and variation in sulcal patterning in the neocortex, the general anatomical positioning of and connectivity between regions is highly stereotyped between individuals, suggesting that a significant proportion of the transcriptional coding for this common architecture is conserved across the human population.”

  • The paper relies on this without support.

    A significant proportion of the transcriptional coding for the common brain architecture is conserved across the human population.

    “Despite a range of brain sizes across individuals and variation in sulcal patterning in the neocortex, the general anatomical positioning of and connectivity between regions is highly stereotyped between individuals, suggesting that a significant proportion of the transcriptional coding for this common architecture is conserved across the human population.”

  • The paper relies on this without support.

    The mouse ISH expression energy values are comparable to human microarray intensities for between-species correlations and network statistics.

    “These values were treated the same way as microarray intensities for between-species correlations and network statistics.”

  • The paper relies on this without support.

    Genes in mouse can be assigned to the same modules as their human orthologs.

    “Genes in mouse were assigned to the same modules as their human orthologs.”

  • The paper relies on this without support.

    The AHBA tissue samples can be mapped to the HCP 52 region parcellation such that gene expression can be compared to functional connectivity.

    “The next step is to map the Allen Human Brain Atlas (AHBA) tissue samples to the HCP 52 region parcellation so that comparison can be made.”
6 of 6

Claims other papers make about this one

Each is checked against this paper’s own text: does it say what the citing paper says it does?

  • High differential stability appears to be a hallmark of genes with considerable biological relevance.

    “high differential stability appears to be a hallmark of genes with considerable biological relevance40.”

    From Oxytocin pathway gene networks in the human brain (2019) · cites this paper as [40]

  • High differential stability is indicative of highly consistent patterning across donor brains.

    “which is indicative of highly consistent patterning across donor brains40.”

    From Oxytocin pathway gene networks in the human brain (2019) · cites this paper as [40]

Self-check

Percentages, sums and reported differences re-computed from the paper’s own numbers.

Each contradiction between two passages listed here was confirmed by an independent check.

  • Mouse regional expression is described as averaged over different voxel sets

    Major

    One methods section says mouse expression energy was averaged across all voxels in each matched structure. Another says it was averaged over only the 250 most highly expressed voxels, except in regions with fewer than 250 voxels.

  • Autworks disease-set count differs between sections

    Minor

    The results give 2,289 Autworks gene sets, while the methods describe the same database as listing 2,288 disease states. The text does not explain the one-set difference.

2 of 2

Cites

Flagged first, then by citations

Research works printed but not linked: 1

  1. Richiardi, J. et al. Correlated gene expression supports synchronous activity in brain networks. Science 348, 1241–1244 (2015). · DOI · Paper page

Cited by

Sorted by citations

Related

Authors
Hawrylycz, M., Miller, J. A., Menon, V., Feng, D., Dolbeare, T., Guillozet-Bongaarts, A. L., Jegga, A. G., Aronow, B. J., Lee, C.-K., Bernard, A., Glasser, M. F., Dierker, D. L., Menche, J., Szafer, A., Collman, F., Grange, P., Berman, K. A., Mihalas, S., Yao, Z., Stewart, L., Barabási, A.-L., Schulkin, J., Phillips, J., Ng, L., Dang, C., Haynor, D. R., Jones, A., Van Essen, D. C., Koch, C., & Lein, E.
Affiliations
  • Allen Institute for Brain Science (US)
  • Cincinnati Children's Hospital Medical Center (US)
  • Washington University in St. Louis (US)
  • Northeastern University (US)
  • Xi’an Jiaotong-Liverpool University (CN)
  • University of Cincinnati (US)
  • University of Washington (US)
  • Georgetown University (US)
Corresponding author
  • Hawrylycz, Michael
  • Lein, Ed
ORCID
Venue
Nature Neuroscience
Publisher
Springer Science and Business Media LLC
Published
2015-11-16
Volume
18
Issue
12
Pages
1832-1844
ISSN
  • 1097-6256 (ISSN-L)
  • 1546-1726
DOI
10.1038/nn.4171
PMC
PMC4700510
OpenAlex
W2260830658
Semantic Scholar
e3aa43b55cdd5d733360a178d938565fa201667a
Cited by
663
References
49
Citations by year
2026: 53 · 2025: 84 · 2024: 83 · 2023: 70 · 2022: 72 · 2021: 74 · 2020: 66 · 2019: 38 · 2018: 46 · 2017: 44 · 2016: 29 · 2015: 1
FWCI
24.04
Type
journal article
Language
en
Open access
Licence
Locations
  • Nature Neuroscience (publishedVersion)
  • PubMed (publishedVersion)
  • Europe PMC (PubMed Central) (acceptedVersion)
  • PubMed Central (acceptedVersion)
  • Digital Access to Scholarship at Harvard (DASH) (Harvard University) (submittedVersion, cc-by)
Funders
  • National Institutes of Health (5r33da027644-04, 5u54mh091657-05)
  • National Institute of Mental Health (1U54 MH091657, U54-MH091657)
  • National Institute on Drug Abuse (R33 DA027644, 4R33DA027644)
Topics
  • Life Sciences › Neuroscience › Cognitive Neuroscience › Functional Brain Connectivity Studies
  • Life Sciences › Biochemistry, Genetics and Molecular Biology › Molecular Biology › Single-cell and spatial transcriptomics
  • Life Sciences › Biochemistry, Genetics and Molecular Biology › Molecular Biology › Bioinformatics and Genomic Networks
Keywords
differential stability, brain cell types, resting-state functional connectivity, neocortex, neurodevelopmental disorders, neurodegenerative diseases, therapeutic target, gene expression signatures
MeSH terms
Adult, Animals, Brain, Humans, Nerve Net, Mice, Gene Regulatory Networks, Transcriptome

About these checks

Registry, retraction, Cites and statistics results are computed, not judged, with one exception: a printed reference that no index lists is read once by a model before it counts as a citation error. Claims are extracted by a model and kept only when their quote appears verbatim in the text; each cited source is then checked, with a model’s help, for whether it says what the claim relies on. Those verdicts can be wrong — read the claim’s quote and the cited source before relying on one.

Claims read by deepseek-v4.1-flash under definition version 3 (extract-v3) on 2026-10-09.

Classified by jev-1.13-20260917 (gold-guideline-v0.2) on 2026-10-09.

Statistics and arithmetic are recomputed, not judged.

Confidence bands are shown only for fields calibrated on our annotated test set. None is calibrated yet, so no band is shown.

What counts as a claim ›How we measure confidence ›How we check this ›

Are you an author, or do you see a problem with this page? Request a review