Transcriptomic Causal Networks identified patterns of differential gene regulation in human brain from Schizophrenia cases versus controls

Akram Yazdani; Michael R Kosorok; Panos Roussos; Raul Mendez-Giraldez

REVIEW 5 major objections 6 minor 10 references

Transcriptomic Causal Networks identified patterns of differential gene regulation in human brain from Schizophrenia cases versus controls

T0 review · 5 major / 6 minor · reviewed 2026-08-14 · deepseek-v4-flash

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗ Review JSON

Pith's one-line read Transcriptomic causal networks built from genetic, RNA, and 3D chromatin data identify nine genes differentially regulated in schizophrenia and four hubs essential for normal brain function.

desk verdict A promising adaptation of G-DAG to brain transcriptomics, but the causal and differential claims outrun the evidence. read the letter →

arxiv 1908.07520 v1 pith:4C2ZVEGO submitted 2019-08-20 q-bio.GN

Akram Yazdani , Raul Mendez-Giraldez , Michael R Kosorok , Panos Roussos This is my paper · ORCID

classification q-bio.GN

keywords schizophreniatranscriptomiccausalnetworkgeneregulationMendelianrandomizationHi-CRNA-seqdifferentialessentialgenes

verification ladder T0 review T1 audit T2 compute T3 formal

The pith

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The reading

Schizophrenia is usually studied one gene at a time, but this paper treats the brain's transcription program as a causal network and asks which regulatory connections change between cases and controls. By combining genetic variants, RNA-seq from post-mortem prefrontal cortex, and 3D chromatin contact data, the authors build directed edges from the principle that genetic variation causes transcription, and identify nine genes whose regulation differs in schizophrenia: GABRA2, LRRTM2, PPM1E, SORT1, GNAL, ZNF69, RALGPS2, ZNF672, and SNRNP48. They also find four hubs, NRXN3, TENM3, MYH10, and PEX5L, that predict downstream transcription in both cases and controls, calling them essential for brain function, and propose five new candidate schizophrenia genes. A sympathetic reader would care because this moves from a list of differentially expressed genes to a causal wiring diagram with specific intervention points.

What carries the argument

The machinery is the transcriptomic causal network, a directed acyclic graph in which an edge from one gene to another means a change in the first gene's transcription causes a change in the second's. Edges are oriented by Mendelian randomization: for each gene, genetic variants are selected from exome-chip SNPs and fetal-brain Hi-C contacts, and because alleles are assigned randomly at conception, variation in the target gene's transcription is treated as a cause, not an effect, of downstream genes. The paper adapts the G-DAG algorithm to build this graph, then classifies genes by in- and out-degree into broadcasters, receptors, and mediators, defines modules around high-out-degree hubs, and tests hub-to-downstream predictions with penalized regression and cross-validation.

What would settle it

A pleiotropy check on the 81 variants with significant transcription effects would settle the edge directions: if a variant assigned to one gene also associates with downstream genes after conditioning on that gene, or if instrument strength is weak (F-statistic below 10), the causal orientation is not identified. A second decisive check is whether adult prefrontal Hi-C contacts reproduce the fetal contacts used to select the instruments.

Watch

Extended reading notes

Core claim

The central claim is that a transcriptomic causal network, built by treating inherited genetic variation as the cause of transcription and using Hi-C contacts to pick cis-acting variants, can separate regulatory wiring that is preserved in schizophrenia from wiring that breaks. Applied to 1,181 transcripts from post-mortem dorsolateral prefrontal cortex, the network yields nine mediator genes whose effect on downstream genes is lost in SCZ cases: GABRA2, LRRTM2, PPM1E, SORT1, GNAL, ZNF69, RALGPS2, ZNF672, and SNRNP48. Three of these, GABRA2, LRRTM2, and PPM1E, are preferentially expressed in brain and related to synaptic function; GNAL sits between two antipsychotic drug targets. Four hubs, NRXN3, TENM3, MYH10, and PEX5L, predict the transcription of their downstream genes in both cases and controls, so the paper labels them essential for brain function; they anchor modules containing many schizophrenia-associated genes. The same logic predicts RNF150, UNC5D, RTF1, LMO7, and SEZ6L as new schizophrenia-associated genes. The paper also observes that highly connected genes are less likely to be influenced directly by genetic variants, implying that schizophrenia risk acts largely through trans-regulation rather than cis effects on hub genes.

Load-bearing premise

Every directed edge assumes the genetic variants assigned to a gene affect that gene's transcription and reach other genes only through it, with no sideways effects and no hidden confounders, and the 3D chromatin contacts used to pick those variants come from fetal brain, not the adult prefrontal cortex being studied.

Editorial extensions

If this is right

The nine differentially regulated genes are candidate causal mediators of schizophrenia-related transcription, not just markers, and are testable in independent post-mortem or stem-cell datasets.
NRXN3, TENM3, MYH10, and PEX5L should be treated as hubs whose disruption would alter many downstream genes, so therapies aimed at downstream targets may avoid breaking essential brain regulation.
RNF150, UNC5D, RTF1, LMO7, and SEZ6L are predicted schizophrenia genes that can be prioritized for targeted sequencing or functional assays.
Because highly connected genes are less likely to carry direct cis-regulatory variant effects, genetic risk for schizophrenia is expected to concentrate in trans-regulation of peripheral genes rather than in hubs themselves.

Reading between the lines

Editorial extensions of the paper, not claims the author makes directly.

Editorial inference: if the instrument-validity assumption fails, the whole edge set is unidentified; a natural next paper would re-build the network with adult prefrontal Hi-C data and report instrument strength statistics.
Editorial inference: the hubs-and-modules structure makes a crisp perturbation prediction: CRISPR interference against NRXN3, TENM3, MYH10, or PEX5L in human neurons should move downstream genes in the signed directions shown, testable without relying on Mendelian randomization assumptions.
Editorial inference: the observation that high-connectivity genes are less cis-regulated may generalize beyond schizophrenia; applying the same pipeline to another brain disorder or tissue would show whether it is a general feature of gene regulatory networks.

Share X Bluesky LinkedIn Reddit HN

Signed reviews

No signed human review yet.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, and a circularity audit.

Desk Editor's Note

read the letter

Colleague,

What you should know about this one: it's the authors' G-DAG method, previously used on metabolomics, applied to schizophrenia postmortem brain. The new bits are the Hi-C-informed variant selection and a case/control comparison of network modules. Nine mediator genes and four hubs are the headline results. The method is interesting, but the paper doesn't support its own strong conclusions.

What it does well: the pipeline integrates genotype, RNA-seq, and Hi-C data into directed networks, which is a step beyond correlation-based gene co-expression. The authors check predictions with cross-validation, and they use external databases (HPA, STRING) to give biological context. That's more than most network papers do.

The soft spots are serious. The causal directions are the foundation of everything, and they rest on Mendelian randomization assumptions that are never tested. We don't get F-statistics or any weak-instrument check for the 81 variants; no pleiotropy screen; and the Hi-C data are from fetal cortex, not adult DLPFC, so the regulatory architecture may not match the expression data. If any variant affects a downstream gene directly or via linkage, the directed edges and everything built on them are spurious.

The differential regulation claim is also underpowered. The "9 genes differentially regulated" come from forest plots with overlapping confidence intervals, with no formal test or multiple-testing correction. That's a central result, and it needs a proper interaction test.

There's also a circular flavor: the network is built on 1,181 transcripts chosen for enrichment of known SCZ-associated genes, and the novel candidates are then predicted from their position relative to those same genes. Conditional analysis is a reasonable way to generate hypotheses, but it's not independent evidence.

Finally, the manuscript refers to a method section that isn't present, and there's no code. That makes it hard to evaluate or reproduce.

So my take: as a hypothesis-generating systems biology paper, it's worth reading. As a claim to have identified causal regulators in schizophrenia, it's not convincing in this form. I'd send it to peer review because the approach and the candidate gene list could be useful to the field, but the authors need to add MR diagnostics, a formal test for differential regulation, and better validation of the novel genes before I'd believe the results.

Referee Report

5 major / 6 minor

Summary. The manuscript proposes an approach that integrates RNA-seq, genotype, and Hi-C data to build "transcriptomic causal networks" via a Mendelian randomization and graphical-modeling framework adapted from the G-DAG algorithm. Applied to CommonMind Consortium dorsolateral prefrontal cortex data, the approach is used to identify differential gene regulatory patterns between schizophrenia cases and controls, to define modules around hubs (TENM3, NRXN3, MYH10, PEX5L) claimed to be essential for brain function, and to propose five novel schizophrenia candidate genes (RNF150, UNC5D, RTF1, LMO7, SEZ6L). The abstract reports nine genes differentially regulated in schizophrenia, including GABRA2, LRRTM2, and PPM1E. The paper is written as a methods-oriented study, but no methods section is included in the provided text.

Significance. If the causal-network estimates and the differential-regulation claims were valid, this would be a substantive contribution to the mechanistic understanding of schizophrenia and to the identification of potential therapeutic targets. The integration of germline genetic variation, gene expression, and chromatin conformation into directed networks is an important direction, and the use of a public, well-characterized dataset is a strength. However, as presented, the central claims are not statistically established: the causal directions rest on untested instrumental-variable assumptions, the differential-regulation results lack formal tests and multiple-testing control, and the module and novel-gene discoveries are conditional on the known schizophrenia-associated gene set used to construct the network. The manuscript also lacks the methodological detail needed for reproducibility.

major comments (5)

[Step 1 (Building transcriptomic causal network) and Results] The causal interpretation of every directed edge rests on the validity of the selected genetic variants as instruments for gene transcription. The manuscript reports only that "81 showed significant effect on gene transcription level" among 4,641 variants, with no F-statistics, no minimum effect-size or p-value threshold, no colocalization with eQTL evidence, and no horizontal-pleiotropy screen. Moreover, the Hi-C contacts used to select variants come from mid-gestation fetal cortex (Won et al., 2016), whereas the RNA-seq data are from adult DLPFC, and regulatory architecture differs across development; a variant that affects multiple genes directly, or is in linkage disequilibrium with another causal variant, would produce spurious directed edges. Because all downstream conclusions, including the nine differentially regulated genes and the four hubs, are built on these edges, the absence of instrument diagnostics is a load-bearing gap.
[Results, "Differential gene regulatory patterns" and Figures 3-4] The central claim that GABRA2, LRRTM2, PPM1E, SORT1, GNAL, ZNF692, ZNF672, SNRNP48, and RALGPS2 are "differentially regulated" in schizophrenia is read from forest plots showing mean effect sizes with 95% confidence intervals for cases versus controls. No formal statistical test of the case-control difference is reported, no p-values accompany the comparisons, and no multiple-testing correction is applied across the nine genes or across the many edges in the network. Overlapping confidence intervals do not establish equivalence, and non-overlap is not a test of significance, so the differential-regulation claim is not statistically supported.
[Step 4 and Results, "Modules" and novel-gene predictions] The module identification and novel-gene discovery are partly circular. The 1,181-gene network is explicitly constructed from "a class of genes that includes the highest number of SCZ-associated genes," and the novel candidates (RNF150, UNC5D, RTF1, LMO7, SEZ6L) are inferred under the assumption "that the SCZ-associated genes are truly related to SCZ." Consequently, the finding that modules are enriched for schizophrenia-associated genes and that novel candidates are connected to known schizophrenia genes is shaped by the prior labels rather than being an independent discovery. No independent replication or external functional validation is provided.
[Methods (missing)] The text repeatedly defers to a "method section" for the G-DAG adaptation, edge p-value cutoffs, degree thresholds, penalization parameters, cross-validation schemes, and conditional-analysis details (Steps 1, 3, and 4; TENM3 module). No such section is present in the manuscript. Without these details, in particular the instrument selection protocol and the structure-learning algorithm, the results cannot be reproduced or independently assessed. This is not purely a presentation issue; it blocks verification of the central causal claims.
[Step 2 (Identification of receptor, broadcaster, and mediator genes)] The classification of genes as broadcasters, receptors, or mediators depends on thresholds for out-degree and in-degree that are described only as "a given out-degree" and "a given in-degree." These thresholds are never specified, and no sensitivity analysis is reported for their choice. Because this classification drives the identification of hubs, mediators, and differential regulatory patterns, the missing criteria and sensitivity analysis constitute a substantive gap.

minor comments (6)

[Abstract and Results] The abstract lists "ZNF69" but the Results section names "ZNF692"; these should be reconciled (the gene symbol appears to be ZNF692, as used in the body text).
[Abstract] "RNAF150" appears in the abstract but the gene is referred to as RNF150 in the Results; please correct the typo.
[Table 1] The module numbering is inconsistent: two rows are both labeled "3" (MYH10 and PEX5L); renumber the modules uniquely.
[NRXN3-Module] The text contains an unresolved reference "[Ref]" when discussing the NRXN3 protein family; this citation needs to be completed.
[MYH10-Module] The text uses the typo "MHYH10" in the discussion of the MYH10 prediction results; this should be corrected to MYH10.
[Supplementary materials] The manuscript references Supplementary Tables 1-13 and Supplementary Figures 1-2, but no supplementary material is provided with the submitted text; please ensure all supplements are included.

Circularity Check

2 steps flagged · score 4.0 of 10

Novel SCZ-gene and module-enrichment claims partially reduce to the input SCZ gene list; the causal network edges themselves are learned from independent genotype-expression data.

self definitional [Step 4 (Methods); Results, 'Modules', TENM3-Module]
"Given the list of SCZ-associated genes that are mapped in our causal network, we found new genes putatively related to SCZ and evaluated our results using conditional analysis assuming that the SCZ-associated genes are truly related to SCZ. All the genes in this pathway are associated with SCZ except RNF150. Therefore, we hypothesized that RNF150 is a SCZ-associated gene as well."

The novel SCZ candidates are not tested for disease association; they are declared candidates because they sit adjacent to genes already on the input SCZ list. For RNF150, the only stated reason is that all other genes in its pathway are SCZ-associated; for UNC5D, the reason is that it directly influences three SCZ-associated genes and is influenced by one. The 'conditional analysis' predicts the candidate's expression from its SCZ-connected regulators, which tests transcriptional regulation, not SCZ association. Thus the output 'novel SCZ-associated' is the input SCZ label propagated through network adjacency, reducing to 'a gene whose neighborhood is enriched for the input list.'
fitted input called prediction [Results, first paragraph; Step 3 (Methods)]
"we identified the transcriptomic network over a set of 1,181 transcripts from a class of genes that includes the highest number of SCZ-associated genes according to Fromer et al. (2016). In order to identify modules related to SCZ, we mapped the SCZ-associated genes, according to liturature, on the transcriptomic network."

The reported 'modules with a high number of SCZ-associated genes' is partly guaranteed by construction: the network's node set was deliberately chosen to maximize SCZ-associated genes, and the modules were then defined by mapping those same SCZ-associated genes onto the network. Any reasonably dense region of this deliberately enriched node set will tend to contain many SCZ genes, so the module-enrichment finding restates the input selection criterion rather than being an emergent data discovery. The directed edges are still learned from genotype and expression data, so only the enrichment claim is circular, not the entire network.

full rationale

The central causal network is not circular: directed edges are oriented by genetic-instrument reasoning (Step 1), and the 9 differentially regulated genes and hub effects are read off the learned network using regression and cross-validation, which are independent of the SCZ label list. Instrument-validity weaknesses (no F-statistics, no pleiotropy screens, fetal rather than adult Hi-C) are correctness risks, not circularity. However, two secondary claims are partially circular. First, the 'novel SCZ-associated genes' (RNF150, UNC5D, RTF1, LMO7, SEZ6L) are produced by propagating the input SCZ-associated gene list through network adjacency; the stated evaluation assumes the known list is true and predicts expression, not disease association, so the novel-gene conclusion reduces to a restatement of the input neighborhood. Second, the module-enrichment finding is shaped by selecting a 1,181-gene node set enriched for SCZ-associated genes before defining modules via those same genes. Because these circularities affect supporting claims rather than the independently learned edge structure, the overall score is moderate.

Assumptions & free parameters 3 free parameters · 4 assumptions · 0 invented entities

The central claims depend on untested causal assumptions (valid genetic instruments, fetal-to-adult Hi-C transferability) and a hand-picked gene set, plus several unstated thresholds; these are not supplied with independent evidence or code.

free parameters (3)

edge p-value cutoff = not specified
In Figures 5C, 6C, 7C, 8C, edges are shown 'above a certain cutoff value' defined as the significance level of the likelihood for an edge; the cutoff is not stated in the text, making the module boundaries unreproducible.
gene set size for network construction = 1,181 transcripts
The network is built on 1,181 transcripts selected as the class of genes with the highest number of SCZ-associated genes (Results, first paragraph). This is a hand-picked subset, not all expressed genes, and it determines which hubs and modules can be found.
degree thresholds for broadcaster/receptor classification = not specified
Genes are classified as broadcaster, receptor, or mediator using 'given out-degree' or 'given in-degree'; the required magnitude of degree is left undefined (Step 2).

assumptions (4)

domain assumption Selected genetic variants act as valid instrumental variables for gene expression: relevant, independent of confounders, and affecting downstream genes only through the instrumented gene (no horizontal pleiotropy).
The G-DAG adaptation rests on Mendelian randomization logic ('genetic inherited variation is the cause of gene transcription not the other way around', Step 1). No instrument strength or pleiotropy checks are reported.
domain assumption Hi-C contacts from mid-gestation developing cerebral cortex represent regulatory interactions in adult dorsolateral prefrontal cortex.
Hi-C libraries are from mid-gestation cerebral cortex (Won et al., 2016), while RNA-seq is from adult DLPFC; the paper applies these contacts to select cis-regulatory variants without validating their adult relevance (Data section).
domain assumption The 1,181-gene subset is sufficient to reconstruct the causal relationships relevant to SCZ.
Genes excluded from the network cannot be hubs, mediators, or receptors, so missing genes could change all inferred pathways (Results, first paragraph).
standard math Bayesian network structure learning (MMHC) yields a consistent DAG under the standard causal Markov and faithfulness assumptions.
The G-DAG algorithm is derived from MMHC (Tsamardinos et al., 2006); these assumptions are standard for constraint-based graph learning.

how reviews work

0 comments

Cite this review

Pith. "Pith review of Transcriptomic Causal Networks identified patterns of differential gene regulation in human brain from Schizophrenia cases versus controls." pith.science (2026). https://pith.science/paper/4C2ZVEGO

@misc{pith2026190807520,
  author       = {Pith},
  title        = {Pith review of: Transcriptomic Causal Networks identified patterns of differential gene regulation in human brain from Schizophrenia cases versus controls},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/4C2ZVEGO}},
  note         = {Machine review of arXiv:1908.07520}
}

read the original abstract

Common and complex traits are the consequence of the interaction and regulation of multiple genes simultaneously, which work in a coordinated way. However, the vast majority of studies focus on the differential expression of one individual gene at a time. Here, we aim to provide insight into the underlying relationships of the genes expressed in the human brain in cases with schizophrenia (SCZ) and controls. We introduced a novel approach to identify differential gene regulatory patterns and identify a set of essential genes in the brain tissue. Our method integrates genetic, transcriptomic, and Hi-C data and generates a transcriptomic-causal network. Employing this approach for analysis of RNA-seq data from CommonMind Consortium, we identified differential regulatory patterns for SCZ cases and control groups to unveil the mechanisms that control the transcription of the genes in the human brain. Our analysis identified modules with a high number of SCZ-associated genes as well as assessing the relationship of the hubs with their down-stream genes in both, cases and controls. In addition, the results identified essential genes for brain function and suggested new genes putatively related to SCZ.

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

10 extracted references · 10 canonical work pages

[1]

Transcriptomic Causal Networks identified patterns of differential gene regulation in human brain from Schizophrenia cases versus controls Akram Yazdani1, Raul Mendez-Giraldez2, Michael R Kosorok3, Panos Roussos1,4,5 1Department of Genetics and Genomic Science, Icahn School of Medicine at Mount Sinai, New York, NY, USA 2Lineberger Comprehensive Cancer Cen...

work page 2009
[3]

in that the latter tend to result in many dependencies that arise from undirected networks and face limited success at finding functional genes (Sullivan, Kendler, & Neale, 2003). However, the causal networks are directed graphs established in the recognition of the hierarchal structure of the biological systems, which provide better understanding of the ...

work page 2003
[5]

the counts matching both exons of a given gene and non-coding RNA according to the newer annotations, could well correspond to missannoated non-coding RNA, with potential trans-regulatory effect over distal genes. For each relevant gene within each of the 5 described modules, we accounted for the level of expression in human brain and other relevant tissu...

work page 2016
[6]

A: Effect of the mediators exclusively expressed in brain on the downstream genes. The forest plot represents 95% confidence intervals of the effect size in downstream gene transcription by the different mediator gene, in red mean effect size for SCZ cases and in blue, mean effect size for controls. B: Upstream genes and downstream genes for the mediators...

work page 2019
[7]

cancer, organismal injury and abnormalities, cellular development

A: TENM3-Module with SCZ-associated genes in cyan. B: Prediction of non-hub gene transcription levels based on transcription of the hub. This plot shows the average Pearson correlation coefficient between observed and predicted values for controls (blue) and SCZ individuals (red). C: Heatmap of the strength of the relationship between genes in the modules...

work page 2013
[8]

Cellular development, cellular Growth and proliferation, nervous system development and function

A: PEX5L-Module with a SCZ-associated in cyan. B: Prediction of non-hub gene transcription levels based on transcription of the hub. The plot shows the average Pearson correlation coefficient between expected and predicted transcription. C: Heatmap of the strength of the relationship between genes in the modules based on the significance of their associat...

work page 2018
[10]

https://doi.org/10.1186/s13073-018-0518-5 Won, H., et al. (2016). Chromosome conformation elucidates regulatory relationships in developing human brain. Nature, 538(7626), 523–527. https://doi.org/10.1038/nature19847 Wright, I. C., et al. (1999). Supra-regional brain systems and the neuropathology of schizophrenia. Cerebral Cortex, 9(4), 366–378. Yazdani,...

work page Pith review arXiv 2016
[2005]

We conclude that genetic variation in the context of schizophrenia works as cis-regulatory elements for genes that are not so much involved in trans-regulatory interactions

(Rex et al., 2010)(Ryu et al., 2006)(Hodges, Newell-Litwa, Asmussen, Vicente-Manzanares, & Horwitz, 2011). We conclude that genetic variation in the context of schizophrenia works as cis-regulatory elements for genes that are not so much involved in trans-regulatory interactions. This property may just be general of gene regulatory networks. Our method pr...

work page 2008

Show all 10 references

[2006]

and test the stability of the network using a backward selection techniques. Step 2: Identification of receptor, broadcaster, and mediator genes The casual network properties measure the genes impacts on the transcription network based on their out-degree and in-degree of a ge...

2017
[2016]

Our method differs from correlation-based networks (He, Chen and Evans, 2007; Wright et al., 1999; Bullmore and Sporns, 2009; Fromer et al.,

for transcriptomic networks, which was originally proposed for metabolomics. Our method differs from correlation-based networks (He, Chen and Evans, 2007; Wright et al., 1999; Bullmore and Sporns, 2009; Fromer et al.,

2007

Pith tools

Reviewed August 14, 2026 · model on record in the stance chip above.

[1] [1]

Transcriptomic Causal Networks identified patterns of differential gene regulation in human brain from Schizophrenia cases versus controls Akram Yazdani1, Raul Mendez-Giraldez2, Michael R Kosorok3, Panos Roussos1,4,5 1Department of Genetics and Genomic Science, Icahn School of Medicine at Mount Sinai, New York, NY, USA 2Lineberger Comprehensive Cancer Cen...

work page 2009

[2] [3]

in that the latter tend to result in many dependencies that arise from undirected networks and face limited success at finding functional genes (Sullivan, Kendler, & Neale, 2003). However, the causal networks are directed graphs established in the recognition of the hierarchal structure of the biological systems, which provide better understanding of the ...

work page 2003

[3] [5]

the counts matching both exons of a given gene and non-coding RNA according to the newer annotations, could well correspond to missannoated non-coding RNA, with potential trans-regulatory effect over distal genes. For each relevant gene within each of the 5 described modules, we accounted for the level of expression in human brain and other relevant tissu...

work page 2016

[4] [6]

A: Effect of the mediators exclusively expressed in brain on the downstream genes. The forest plot represents 95% confidence intervals of the effect size in downstream gene transcription by the different mediator gene, in red mean effect size for SCZ cases and in blue, mean effect size for controls. B: Upstream genes and downstream genes for the mediators...

work page 2019

[5] [7]

cancer, organismal injury and abnormalities, cellular development

A: TENM3-Module with SCZ-associated genes in cyan. B: Prediction of non-hub gene transcription levels based on transcription of the hub. This plot shows the average Pearson correlation coefficient between observed and predicted values for controls (blue) and SCZ individuals (red). C: Heatmap of the strength of the relationship between genes in the modules...

work page 2013

[6] [8]

Cellular development, cellular Growth and proliferation, nervous system development and function

A: PEX5L-Module with a SCZ-associated in cyan. B: Prediction of non-hub gene transcription levels based on transcription of the hub. The plot shows the average Pearson correlation coefficient between expected and predicted transcription. C: Heatmap of the strength of the relationship between genes in the modules based on the significance of their associat...

work page 2018

[7] [10]

https://doi.org/10.1186/s13073-018-0518-5 Won, H., et al. (2016). Chromosome conformation elucidates regulatory relationships in developing human brain. Nature, 538(7626), 523–527. https://doi.org/10.1038/nature19847 Wright, I. C., et al. (1999). Supra-regional brain systems and the neuropathology of schizophrenia. Cerebral Cortex, 9(4), 366–378. Yazdani,...

work page Pith review arXiv 2016

[8] [2005]

We conclude that genetic variation in the context of schizophrenia works as cis-regulatory elements for genes that are not so much involved in trans-regulatory interactions

(Rex et al., 2010)(Ryu et al., 2006)(Hodges, Newell-Litwa, Asmussen, Vicente-Manzanares, & Horwitz, 2011). We conclude that genetic variation in the context of schizophrenia works as cis-regulatory elements for genes that are not so much involved in trans-regulatory interactions. This property may just be general of gene regulatory networks. Our method pr...

work page 2008