Single Nucleotide Polymorphisms in ANK3 and Psychiatric Risk: A Meta-Analysis, Systematic Review, and Quantitative Trait Locus Insights

TITLE & ABSTRACT

 Summary Assessment

The title is comprehensive and accurately reflects the study scope, incorporating three key methodological components (meta-analysis, systematic review, QTL). The abstract is well-structured with distinct Background, Objectives, Methods, Results, and Conclusion sections; however, specific methodological detail is limited, and key statistical parameters lack precision in reporting.

 Major Issues

Issue 1: Abstract reporting lacks methodological specificity
- Description: The abstract states "A meta-analysis and systematic literature review were conducted" but omits critical methodological details essential for evaluating reproducibility: databases searched, search date, number of studies included in meta-analysis, inclusion/exclusion criteria summary, and specific genetic models analyzed (allelic, dominant, recessive, additive).
- Location: Abstract, Methods section (lines omitted in pagination)
- Impact on Validity: Readers cannot assess completeness of evidence synthesis or replicate the search strategy; may underestimate heterogeneity or methodological rigor.
- Required Author Action: Expand abstract Methods to include: (1) databases and date range; (2) number of studies analyzed in meta-analysis vs. synthetic review; (3) study design types; (4) genetic models assessed (mandatory for genetic association meta-analyses per PRISMA 2020 item S4); (5) population stratification approach.

Issue 2: Abstract Results section lacks effect estimate precision and comparative context
- Description: Multiple SNP-disease associations are reported with ORs and P-values, but the abstract conflates highly significant associations (rs1938526: P = 7.26×10⁻¹²) with nominally significant findings (three PTSD SNPs: P = 0.045). No mention of which findings met the FDR correction threshold vs. nominal significance only, creating ambiguity about robust vs. suggestive signals.
- Location: Abstract, Results section
- Impact on Validity: Readers may overinterpret nominal findings as definitively associated; the abstract does not clearly distinguish between FDR-significant (sQTL findings) and nominal-only associations (eQTL, apaQTL).
- Required Author Action: Clarify abstract Results to distinguish effect tiers: (1) FDR-significant meta-analyses; (2) FDR-significant QTL findings; (3) nominal-only findings. Specify that PTSD SNPs are based on single studies or small sample meta-analyses.

Issue 3: Abstract Keywords underrepresent study scope
- Description: Keywords list "ANK3, psychiatric disorders, bipolar disorder, schizophrenia, major depression, post-traumatic stress disorder, single nucleotide polymorphism, QTL analysis" but omits critical descriptors: "meta-analysis," "systematic review," "gene expression," "alternative splicing," "population stratification," "Hardy-Weinberg equilibrium."
- Location: Keywords section
- Impact on Validity: Reduced discoverability and indexing specificity in PubMed/Web of Science; readers searching for meta-analysis methodology or gene expression QTL studies may miss this work.
- Required Author Action: Expand keywords to 8–12 terms: Add "meta-analysis," "systematic review," "eQTL," "splicing QTL," "population stratification," "Hardy-Weinberg equilibrium." Consider adding "psychiatric genetics" or "neuropsychiatric risk."

 Minor Issues & Editorial Suggestions

1. Title hyphenation: Title reads "Single Nucleotide Polymorphisms in ANK3 and Psychiatric Risk: A Meta-Analysis, Systematic Review, and Quantitative Trait Locus Insights." For enhanced clarity and brevity, consider: "ANK3 Genetic Variants and Psychiatric Disorder Risk: Meta-Analysis, Systematic Review, and Functional QTL Characterization."

2. Abstract structure clarity: Separate "Objectives" as a distinct paragraph; currently merged with Background, reducing clarity.

 Suggested Additional References

For strengthening abstract methodology reporting, consider citing:
- Moher D, Shamseer L, Clarke M, et al. Preferred Reporting Items for Systematic Review and Meta-Analysis Protocols (PRISMA-P) 2015 statement. Syst Rev. 2015;4:1. [PRISMA 2020 extension for transparent reporting of genetic association meta-analyses]
- Stroup DF, Berlin JA, Moher D, et al. Meta-analysis of Observational Studies in Epidemiology (MOOSE) guidelines. JAMA. 2000;283(15):2008–2012. [Standard for observational genetic studies]



INTRODUCTION

 Summary Assessment

The Introduction is well-motivated and clearly establishes the biological significance of ANK3 in neuronal signaling and its implicated role in multiple psychiatric disorders. The rationale for systematic integration of meta-analysis with QTL analysis is compelling. However, the section lacks an explicit statement of study objectives, and the transition between motivating background and study-specific aims could be more precise. PRISMA 2020 item 5 (PICO specification) is incomplete.

 Major Issues

Issue 1: Study Objectives and PICO framework are not explicitly defined
- Description: The Introduction states "To address this gap, the present study integrates meta-analysis with QTL-based transcriptomic analysis..." but does not formally articulate the study's specific objectives in structured PICO format (Population, Intervention/Exposure, Comparison, Outcome). PRISMA 2020 mandates explicit PICO specification in the Introduction or Methods.
- Location: Introduction, final paragraphs (page 3)
- Impact on Validity: Without an explicit PICO definition, readers cannot determine whether specific population strata, diagnostic criteria, or outcome measures were predefined or post-hoc. Critical for assessing risk of selective outcome reporting.
- Required Author Action: Add explicit PICO statement before the Methods section or at the end of Introduction:
  - Population: Individuals (case and control) with psychiatric disorder diagnoses (bipolar disorder, schizophrenia, major depressive disorder, post-traumatic stress disorder); across diverse populations (European, North American, Asian).
  - Exposure: ANK3 single-nucleotide polymorphisms (SNPs); genetic variants in intronic or exonic regions.
  - Comparison: Case vs. control groups matched by population.
  - Outcomes: Primary: odds ratio (OR) of psychiatric disorder risk per ANK3 SNP; Secondary: tissue-specific regulatory effects (eQTL, apaQTL, sQTL) in brain tissues.

Issue 2: Insufficient detail on prior inconsistencies in GWAS findings
- Description: The Introduction notes "considerable inconsistencies remain regarding which ANK3 SNPs are most robustly associated with psychiatric disorders and whether these associations are consistent across diverse populations," but does not cite specific conflicting findings or quantify the extent of replication failure.
- Location: Introduction, page 2, paragraph 4
- Impact on Validity: Readers cannot assess whether prior conflicting results reflect genuine population stratification, genotyping artifact, study quality variation, or statistical chance. Limits the understanding of study significance.
- Required Author Action: Insert a table (or extended text) summarizing 3–5 prior ANK3 GWAS findings showing conflicting results, e.g., "rs10994336 was significant for bipolar disorder in European cohorts (Ferreira et al., 2008; Scott et al., 2009) but non-significant in East Asian populations (Takata et al., 2011; Lee et al., 2011), suggesting possible population-specific genetic architecture." This grounds the need for population-stratified meta-analysis.

Issue 3: Alternative splicing and ANK3 isoforms are not discussed in the Introduction
- Description: The Introduction briefly mentions ANK3 isoforms ("expressed as multiple isoforms arising from alternative splicing," page 2) but does not elaborate on isoform-specific subcellular localization or functional differences, limiting reader understanding of why splicing QTL findings are biologically critical.
- Location: Introduction, pages 2–3
- Impact on Validity: Readers unfamiliar with ANK3 biology will not appreciate the mechanistic importance of the sQTL results presented in the Results section (particularly regarding exon 37 inclusion affecting AIS localization). This weakens the Discussion's functional interpretation.
- Required Author Action: Expand the isoform description: "ANK3 encodes multiple protein isoforms ranging from ~190 kDa (postsynaptic form) to 480 kDa (AIS-localized form), differing in their N-terminal and C-terminal regions. The larger isoforms are critical for axon initial segment (AIS) organization, clustering voltage-gated sodium channels (Nav), and cell adhesion molecules (βIV-spectrin, neurofascin) to stabilize neuronal polarity and action potential generation. Loss-of-function mutations or splice variants affecting AIS-localized isoforms compromise neuronal excitability and have been implicated in neurodevelopmental and psychiatric phenotypes."

 Minor Issues & Editorial Suggestions

1. Reference formatting inconsistency: References are cited numerically (e.g., "ANK3 variants have been implicated in several neuropsychiatric disorders.⁴") but several lack complete bibliographic details in the main text. Verify that all numbered citations in the Introduction correspond to entries in the References section (cross-check with Ref. 1–7).

2. Terminology precision: The term "noncoding regulatory elements" (page 3) is broad and could refer to enhancers, silencers, miRNA binding sites, or splicing regulatory elements. Consider specifying: "intronic enhancer elements or splicing regulatory motifs" based on QTL findings.

3. QTL methodology preview: The Introduction briefly mentions "expression quantitative trait locus (eQTL)," "alternative polyadenylation QTL (apaQTL)," and "splice QTL (sQTL)" analyses, but does not explain what these measure or why they are complementary. A 2–3 sentence explanation would strengthen reader comprehension, e.g., "eQTL studies quantify the effect of genetic variants on steady-state gene expression levels; sQTL analyses assess effects on splicing patterns; apaQTL studies measure effects on alternative transcript 3′-end polyadenylation. Together, these approaches characterize both quantitative and qualitative transcriptomic impacts of disease-associated variants."



MATERIALS & METHODS

 Summary Assessment

The Methods section provides a reasonable structural overview of the meta-analysis workflow and QTL analysis approach; however, critical methodological details are incomplete or imprecisely reported. Search strategy is inadequately documented, risk-of-bias assessment lacks detail, heterogeneity assumptions are not justified, and statistical model choices (fixed vs. random effects) are not explained. PRISMA 2020 items 6–11 are partially compliant.

 Major Issues

Issue 1: Literature search strategy is incompletely reported (PRISMA 2020 item 7 non-compliant)
- Description: The Methods states "two independent researchers (TG and YW) systematically searched PubMed, Web of Science, and MEDLINE using combinations of the terms 'genome-wide association study' with 'ANK3', as well as 'nucleotide polymorphisms' with 'ankyrin 3' or 'Ankyrin-G'." No Boolean operators specified, no search date provided, no search strategy documentation (e.g., MeSH terms, truncation, field tags), and no justification for the specific database selection or exclusion of other major databases (e.g., Embase, Cochrane Central Register of Controlled Trials).
- Location: Methods, "Literature retrieval and selection" section, page 3
- Impact on Validity: Readers cannot reproduce the search, assess whether important studies were missed, or determine if publication bias is likely (missing studies in non-indexed sources). Transparency is compromised; reproducibility cannot be independently verified.
- Required Author Action: Provide a detailed search strategy meeting PRISMA 2020 item 7:
  - Specify exact date range of searches (e.g., "searches conducted from 1 January 2000 through 30 September 2025")
  - Report complete search strings for each database:
    - PubMed: `("ANK3" OR "ankyrin 3" OR "ankyrin-3" OR "ankyrin-G" OR "ANKRD4") AND ("single nucleotide polymorphism" OR "SNP" OR "rs[0-9]{7}" OR "polymorphism" OR "genotype" OR "allele") AND ("bipolar disorder" OR "schizophrenia" OR "depression" OR "PTSD" OR "psychiatric disorder")`
    - Web of Science: [with appropriate field tags: TS= (topic search)]
    - EMBASE (if not already searched): [with EMTREE keywords and field codes]
  - Justify database selection (e.g., "PubMed for broad coverage of life sciences, Web of Science for citation tracking and multidisciplinary coverage, MEDLINE (NLM) indexing")
  - Document supplementary searches (reference harvesting, contact with authors, trial registry searches)
  - Provide PRISMA-S (Search) reporting form documenting search filters applied (peer-review status, publication language, study design filters)

Issue 2: Newcastle-Ottawa Scale assessment is inadequately reported; itemized scoring is unavailable
- Description: Methods mention "The studies with controlled trials were evaluated by the Newcastle-Ottawa Scale (NOS)⁸'⁹ by two different researchers independently. NOS includes 1. Selection... [partial list]... The total score is 9, and if a study can get over 5, it can be assessed as good quality. Those articles that did not meet the quality requirement were excluded." However, no itemized NOS scores are reported for each included study, precluding verification of quality assessments and identification of potential quality-based bias. The threshold of ≥5/9 is not justified.
- Location: Methods, "Quality assessment of included studies" section, page 3
- Impact on Validity: High risk of quality bias; readers cannot assess whether lower-quality studies were differentially included/excluded or whether risk-of-bias patterns differ between psychiatric disorders. Inability to conduct sensitivity analysis by study quality.
- Required Author Action:
  - Create supplementary table (Appendix or Extended data) listing all 27 included studies with complete NOS itemized scores across all 8 categories (Selection: 4 items; Comparability: 1 item; Outcome: 3 items) and total score
  - Justify ≥5/9 threshold with reference to NLM/Cochrane guidance (note: NOS maximum is typically 9 points; confirm scoring system used)
  - Conduct sensitivity analysis excluding low-quality studies (NOS <5) and report whether meta-analysis effect estimates are stable
  - Report inter-rater agreement (Cohen's κ) between TG and YW for NOS assessments
  - Disclose whether specific NOS domains (e.g., Selection of cases, Selection of controls) correlated with effect size heterogeneity

Issue 3: Meta-analysis model selection is not justified; fixed vs. random effects are not discussed
- Description: Methods state: "Utilizing the 'meta (version 4.18-0)' package within R 4.0.1 and RStudio, we conducted a meta-analysis on the specified SNPs. Odds ratios (OR) and 95% confidence intervals (95%CI) were input to yield synthesized outcomes." No specification of whether fixed-effect or random-effects models were used, no justification for model choice, and no reporting of I² heterogeneity statistics or τ² (tau-squared) estimates in the Methods (though these appear in Results).
- Location: Methods, "Synthesis and analysis of data and assessment of publication bias" section, page 4
- Impact on Validity: Model choice affects confidence intervals and P-values critically; random-effects models produce wider CIs and lower P-values than fixed-effect models. Absence of pre-specification suggests possible post-hoc model selection driven by desired results. Standard guidance (Cochrane Handbook 2021) recommends random-effects for anticipated heterogeneity.
- Required Author Action: Add an explicit statement to Methods:
  - "We employed random-effects meta-analysis models using restricted maximum likelihood (REML) estimation as default, as we anticipated heterogeneity across studies due to differences in population ancestry, study design, diagnostic criteria, and genotyping platforms. For each SNP-disorder association with fewer than 5 studies (e.g., rs139972937 for bipolar disorder), we reported individual study estimates and effect directions but did not conduct quantitative meta-analysis due to limited power to estimate τ². Heterogeneity was quantified using I² (percentage of variance due to heterogeneity) and τ² (between-study variance). We report 95% prediction intervals (PI) to characterize the range of effects expected in future populations. If I² ≤50%, we also report fixed-effect estimates for comparison."
  - Specify software defaults: If using R meta package, document whether tau.common=FALSE (default REML)
  - Clarify whether genetic models (allelic, dominant, recessive) were specified a priori or determined post-hoc based on data fit

Issue 4: Handling of genetic models, allele coding, and population stratification is not transparent
- Description: Methods state "Risk estimates were extracted in the form of odds ratios (OR), while SNP relationships were assessed using either minor allele frequency (MAF) or specific genotype distributions. For cohorts stratified by region, meta-analysis was performed only when a consistent direction of effect was observed across all subgroups." However, the Methods does not specify: (a) which genetic models were extracted/analyzed for each SNP (allelic, dominant, recessive, additive/genotype contrast), (b) how studies reporting different genetic models were harmonized, (c) how studies reporting MAF-based analysis were converted to genotype-based OR, and (d) what constitutes "consistent direction of effect" (e.g., all subgroups protective, all subgroups risk-increasing).
- Location: Methods, "Criteria for inclusion" and "Synthesis and analysis of data" sections, pages 3–4
- Impact on Validity: Standard genetic association best practices (including PLINK, GWAS-MA guidelines) mandate transparency in genetic model specification to avoid selective reporting of significant models. Pooling different genetic models without documentation risks bias. Population stratification heterogeneity may reflect genuine allelic heterogeneity vs. linkage disequilibrium differences; undocumented handling obscures this.
- Required Author Action:
  - Create supplementary table extracting for each SNP-disorder pair: (1) genetic models reported (allelic, dominant, recessive, additive), (2) per-study effect estimates for each model if available, (3) decision criteria for selecting primary model for meta-analysis
  - Explicitly define "consistent direction of effect": e.g., "All subgroups with P<0.05 showed protective effects (OR<1)" vs. "≥75% of subgroups showed the same direction, even if nominal significance varied."
  - For studies reporting MAF frequencies and case/control counts, show the conversion formula: e.g., "OR = [case_MAF/(1-case_MAF)] / [control_MAF/(1-control_MAF)]."
  - Report whether the Hardy-Weinberg equilibrium (HWE) was assessed for each study and population separately. In Methods, clarify HWE threshold used (p<0.05, <0.01, <0.001) and whether studies with HWE violation were excluded or separately analyzed (currently only mentioned in Results for rs10761482)

Issue 5: Publication bias assessment rationale inadequately justified
- Description: Methods state: "We combined the funnel plot and the Egger test¹¹ to assess the publication bias of the meta-analysis. As none of the meta-analyses included 10 or more studies, publication bias was not assessed, in accordance with Cochrane guidance. This is internally contradictory: the statement says publication bias "was not assessed," but earlier claims funnel plot and Egger test "were combined." Additionally, the Cochrane Handbook 2021 states that publication bias assessment is unreliable with <10 studies, but does not recommend abandoning it; instead, contour-enhanced funnel plots and interpretation with caution are advised.
- Location: Methods, "Synthesis and analysis of data and assessment of publication bias" section, page 4
- Impact on Validity: Undermines assessment of potential bias in effect estimates; small study effects (which may reflect publication bias, selective outcome reporting, or true effect heterogeneity) are not characterized. Readers cannot determine whether reported associations may be inflated due to the selective publication of positive findings.
- Required Author Action:
  - Clarify that funnel plots were generated for all meta-analyses (even if <10 studies) to visually assess asymmetry
  - Report Egger regression test results for all SNP-disorder combinations (provide slope, intercept, P-value, I²)
  - Justify the post-hoc decision not to assess publication bias: cite Cochrane Handbook p. XXX or alternative guidance; alternatively, report results with statement of statistical limitations
  - Consider alternative publication bias approaches suitable for small samples: trim-and-fill method, or Begg's rank correlation test (less sensitive to heterogeneity than Egger test)
  - Create visual funnel plots for Figures/Supplementary Materials

Issue 6: QTL analysis methodology lacks critical details on multiple testing correction and interpretation thresholds
- Description: Methods state: "To correct for multiple testing across tissues and regulatory layers, we applied Benjamini–Hochberg false discovery rate (FDR) correction to the nominal p-values within each QTL modality (eQTL, apaQTL, and sQTL). Associations with FDR < 0.05 were considered statistically significant." However, several important details are missing: (1) How many tissues and cell types were tested per SNP? (2) What constitutes a "tissue" for FDR correction—are brain regions (neocortex, cerebellum, hippocampus, etc.) and cell types (neurons, astrocytes, microglia) counted separately? (3) Was the FDR correction applied within each QTL type separately (as stated) or across all three modalities combined? (4) Nominal P<0.05 is mentioned as a threshold for interpretation of suggestive findings, but no justification is provided for this threshold.
- Location: Methods, "QTL analysis" section, page 4
- Impact on Validity: Transparency of multiple-testing correction directly impacts the reliability of findings. If FDR correction was lenient (applied only within each tissue, not across tissues), more associations would appear significant; if stringent (applied across all tissues and all three modalities), fewer would. The Discussion acknowledges that eQTL and apaQTL signals "did not remain significant following false discovery rate (FDR) correction," suggesting correction was stringent. However, readers cannot assess whether the sQTL FDR=0.010 results are robust.
- Required Author Action:
  - State total number of tests: "We tested 10 SNPs × N tissues × 3 regulatory modalities = XXX tests total. FDR correction was applied independently within each regulatory modality (eQTL, apaQTL, sQTL), yielding XXX independent testing families. For eQTL, we tested across XXX tissues/cell types; for apaQTL across XXX tissues; for sQTL across XXX tissues."
  - Define tissues explicitly: list all included tissues/cell types (e.g., "neocortex, neurons, microglia, cerebellum, putamen, hippocampus, prefrontal cortex, amygdala" or cite Supplementary Table)
  - Justify FDR threshold: "FDR<0.05 was chosen to control the expected proportion of false discoveries at 5%, balancing sensitivity and specificity given the focused, hypothesis-driven design of our analysis (testing 10 pre-selected SNPs identified through meta-analysis, not genome-wide scan)."
  - For nominally significant but FDR-nonsignificant findings (eQTL, apaQTL), report effect sizes and consistency across tissues to supportthe  biological plausibility claim in the Discussion

Issue 7: Python and Bash script details are insufficient for reproducibility
- Description: Methods state: "All datasets were parsed and harmonised by rsID using custom Python 3.11 and Bash scripts." No mention of script availability, algorithm details, or validation procedures. This is critical for QTL data curation.
- Location: Methods, "QTL analysis" section, page 4
- Impact on Validity: Others cannot verify data parsing accuracy, assess whether variants were correctly matched across databases, or reproduce the analysis. Potential for systematic data misalignment errors.
- Required Author Action:
  - Provide GitHub repository or Figshare code deposit with annotated scripts and documentation
  - Describe parsing algorithm: e.g., "SNPs were matched across eQTL Catalogue and GTEx using exact rsID match (e.g., rs10761482). For multi-allelic SNPs, we retained the major/minor allele designation from GTEx v10 and verified consistency with eQTL Catalogue. X in 10, XXX SNPs could not be uniquely matched and were excluded with details in Supplementary Table X."
  - Report validation: e.g., "Randomly selected 50 SNPs (~0.5%) were manually verified against source databases; 100% concordance achieved"

 Minor Issues & Editorial Suggestions

1. Criteria for exclusion clarity: Exclusion criteria state "Studies employing hazard ratios (HR) for SNP risk estimation were excluded" without justification. HRs are less common for SNP-disease associations but may appear in prospective cohort studies. Consider clarifying whether prospective study designs were excluded or only HR estimates; if the latter, should HR studies be converted to OR for meta-analysis?

2. Diagnostic criteria specificity: Methods do not specify which diagnostic criteria were required (DSM-5, ICD-10, expert-assessed, self-reported). This affects phenotype harmonization; psychiatric disorders have evolved diagnostic definitions. Recommend: "Included studies were required to use standardized diagnostic criteria (DSM-III-R, DSM-IV, DSM-5, or ICD-10) and to provide data allowing assessment of case definition clarity."

3. Control group definition: Methods state "For each patient cohort, a matched control group comprising healthy individuals from the same population and study was selected," but do not specify: Were control groups screened for absence of psychiatric disorders (negative screening) or merely reported as healthy/normal? This affects the interpretation of effect estimates.

4. Extended Dataset sourcing: Methods reference "Extended dataset 1 (Figshare, https://doi.org/10.6084/m9.figshare.29886620)" but Figshare URLs are not stable; recommend archiving on an institutional repository with a persistent identifier. Verify accessibility at submission time.

5. Gene structure sourcing: Methods cite Ensembl database for exon/intron mapping: "Intron and exon details for human ANK3 were sourced from the Ensembl protein database." Specify Ensembl release version (e.g., Ensembl release 114, GRCh38.p14) for reproducibility and to account for future genome builds.

6. Statistical software citation: Citation of "meta (version 4.18-0)" package is appropriate; however, cite the R package publication: Schwarzer G. meta: An R package for meta-analysis. R News. 2007;7(3):40–45. Similarly, specify GraphPad Prism version and matplotlib version.

RESULTS
Summary Assessment

The Results section is well-organized, presenting study selection flow, quality assessment, meta-analysis findings by psychiatric disorder, SNP localization, and QTL analyses. However, critical numerical details are missing: forest plots lack heterogeneity statistics (I², τ²), effect estimates for some SNPs/populations are incomplete, and QTL tissue selection is not justified. Data tables are presented, but summaries lack precision.

 Major Issues

Issue 1: Heterogeneity statistics (I², τ²) not reported for meta-analysis results
- Description: Results section reports meta-analysis point estimates and P-values (e.g., "rs10994336... emerged as a significant risk locus (OR = 1.28, 95% CI: 1.13–1.45, P = 1.04×10⁻⁴)") but does not report I² (proportion of variance due to heterogeneity) or τ² (between-study variance) for any SNP-disorder combination. Figures 2–5 show forest plots but lack visual or tabular heterogeneity reporting. This is a standard requirement of meta-analysis reporting (PRISMA 2020 item 20).
- Location: Results section, pages 5–10 (Meta-analysis sections for each disorder)
- Impact on Validity: Readers cannot assess the degree of heterogeneity, identify whether the fixed-effect model assumption (homogeneity) is justified, or determine whether effect estimates are generalizable across populations. Large heterogeneity suggests true effect variation due to population genetics, diagnostic criteria, or genotyping methods; small heterogeneity suggests consistent effects. This distinction is critical for interpretation.
- Required Author Action:
  - For each SNP-disorder combination analyzed quantitatively, report: I², 95% CI for I² (if available from meta package), τ², Q-test P-value
  - Add column to Results tables (if created) or supplementary table with heterogeneity statistics
  - For forest plots (Figures 2–5, 6), add text overlay showing I² and P(heterogeneity), or create a companion table
  - Example revision for rs10994336 in bipolar disorder: "Rs10994336, included in nine studies from European and North American populations, emerged as a significant risk locus (OR = 1.28, 95% CI: 1.13–1.45, P = 1.04×10⁻⁴, I² = XX%, τ² = XX). Heterogeneity was [low/moderate/high], suggesting [consistent/variable] effects across populations."

Issue 2: Study selection Flow diagram (Figure 1) lacks detail; numbers do not reconcile
- Description: Figure 1 shows the literature screening process: "1,140 articles retrieved → [screening] → Full-text review resulted in exclusion of 214 studies → 27 studies included." However, the flow diagram does not show intermediate steps: How many articles were excluded during title/abstract screening? How many underwent full-text review? The caption refers to "For exclusion, please refer to section Criteria for Exclusion," but does not quantify exclusion reasons, which is a PRISMA 2020 item 13 requirement.
- Location: Results, page 5, Figure 1
- Impact on Validity: Readers cannot assess whether screening was systematic or selective. For example, if 900 of 1,140 articles were excluded during abstract screening (likely given search breadth), were they appropriately excluded, or did reviewers prematurely filter? Without exclusion reason tallies, the risk of selection bias cannot be evaluated. PRISMA 2020 mandates a flow diagram showing: (1) Total records identified through searching, (2) duplicate records excluded, (3) records screened, (4) records excluded, (5) full-text articles assessed, (6) reasons for exclusion at full-text stage, (7) studies included in review.
- Required Author Action:
  - Revise Figure 1 to include:
    - n= records identified through database searching (e.g., PubMed: n=XXX; Web of Science: n=XXX; MEDLINE: n=XXX)
    - n= duplicates excluded
    - n= records screened (after duplicate removal)
    - n= records excluded during title/abstract screening [with top 3–5 exclusion reasons and counts]
    - n= full-text articles assessed for eligibility
    - n= studies excluded at full-text stage [itemized: e.g., "n= without ANK3 genotype data, n= non-psychiatric outcomes, n= case reports/reviews, etc."]
    - n= studies included in meta-analysis
    - n= studies included in qualitative synthesis
  - Create Supplementary Table detailing exclusion reasons for all 214 full-text rejections (at minimum, top reasons and counts)

Issue 3: Subgroup analysis rationale for rs10761482 HWE stratification is not clearly justified in the Results
- Description: Results state: "To further explore potential population-specific effects, subgroup analysis was conducted for rs10761482 within the Asian population, which was stratified into three subgroups based on Hardy–Weinberg equilibrium (HWE) status and ethnicity. The first subgroup comprised Chinese cohorts not in HWE..., the second subgroup included Iranian participants, and the third subgroup consisted of Chinese cohorts with significant HWE conformity. However, the Results do not explain: (a) What was the predetermined HWE P-value threshold used to classify "in HWE" vs. "not in HWE"? (b) Why were Chinese studies stratified by HWE while Iranian studies were not? (c) How many studies fell into each subgroup (n=?), and were cell counts adequate for meta-analysis? This subgroup decision appears post-hoc, driven by observing HWE violations in data, increasing the  risk of cherry-picking.
- Location: Results, page 6, Schizophrenia section
- Impact on Validity: If HWE stratification was not pre-specified, this subgroup analysis inflates Type I error and may reflect data-driven hypothesis generation. The Discussion mentions HWE issues (pages 11–12) but does not clearly state whether the HWE analysis was predefined.
- Required Author Action:
  - Clarify in Results: "HWE status was assessed a priori for each included study using the standard χ² goodness-of-fit test (threshold P < 0.05 for HWE deviation). Studies significantly deviating from HWE were stratified separately in subgroup meta-analyses, as recommended by [citation], to investigate potential sources of heterogeneity (e.g., population substructure, genotyping error, selection bias)."
  - Report complete stratification: "For rs10761482 in schizophrenia, n=X Chinese studies with HWE conformity (P≥0.05) were meta-analyzed separately; n=Y Chinese studies with HWE deviation (P<0.05) were analyzed separately; n=Z Iranian studies were analyzed separately (HWE status: [report]). Overall meta-analysis combining all Asian studies: [results]. Subgroup-by-ethnicity interaction: [report I² and τ² for between-subgroup heterogeneity]."

Issue 4: PTSD SNP associations based on inadequate sample sizes (single studies or n=3 total)
- Description: Results state: "For PTSD, all three investigated variants—rs28932171, rs11599164, and rs17208576—were found to be significantly associated with reduced risk, each exhibiting the same odds ratio (OR = 0.48) and a nominally significant P-value (P = 0.045)." However, Extended Dataset 3 (not fully visible in the manuscript) likely shows that these three SNPs were reported in only 1–2 studies; meta-analysis with 1 study is meaningless (it is simply the study estimate), and meta-analysis with 2–3 small studies has minimal power. The identical OR=0.48 and P=0.045 for all three variants is also suspicious and suggests these may be from a single study or linkage disequilibrium. This finding is misrepresented in the Abstract as a major result.
- Location: Results, page 7, section "Meta-analysis and synthetic review... PTSD"
- Impact on Validity: Readers may over-interpret these PTSD associations as robust findings when they are actually nominal signals from single studies or small samples. No sensitivity analysis or replication evidence is presented. This weakens the overall conclusions.
- Required Author Action:
  - In Results, explicitly report sample sizes per SNP-disorder pair: "For PTSD, meta-analysis of rs28932171 included n=X independent studies (total cases=XXX, controls=XXX); rs11599164 included n=Y studies; rs17208576 included n=Z studies. [Note: If n<3 studies, these are not true meta-analyses but rather synthetic reviews of single or paired studies.]"
  - Explain identical OR and P-values across three SNPs: "These three SNPs were derived from [1 study / 2 studies] (authors: [citation]). The identical effect estimates reflect their origin in a single study population and should be interpreted with caution; replication in independent cohorts is required."
  - In Discussion, downgrade PTSD findings to "preliminary" or "suggestive" pending replication

Issue 5: QTL analysis, tissue selection, and justification are absent
- Description: QTL analyses state: "eQTL slopes and p values were retrieved from the eQTL Catalogue... apaQTL (alternative polyadenylation) and sQTL (splicing) summary statistics were retrieved from the GTEx v10 resource... For every brain tissue, we extracted the slope (β), its standard error, and the nominal p value." However, which specific tissues and cell types were included is not stated in Methods or Results. For example, did the analysis include all available eQTL Catalogue tissues, or was a priori selection (e.g., psychiatric-relevant tissues: prefrontal cortex, amygdala, nucleus accumbens, hippocampus, anterior cingulate cortex) applied? The eQTL Catalogue contains diverse tissues; arbitrary inclusion may inflate the multiple testing burden and reduce power.
- Location: Results, pages 8–11, QTL analysis sections; Methods page 4
- Impact on Validity: If all available tissues were analyzed without selection, the FDR correction may have been too stringent, and true signals in relevant tissues may be missed. Conversely, if tissue selection was post-hoc (motivated by observing significant results), this increases false discovery risk.
- Required Author Action:
  - Provide supplementary table listing: (1) All tissues analyzed for eQTL; (2) All tissues analyzed for apaQTL; (3) All tissues analyzed for sQTL
  - State selection criteria: "We extracted data for all available tissues in the eQTL Catalogue and GTEx v10 to comprehensively assess regulatory effects. No a priori tissue selection was performed. FDR correction was applied within each regulatory modality across all tested tissues, accounting for multiple testing."
  - OR if tissue selection was applied: "We prioritized brain tissues implicated in psychiatric disorder etiology: prefrontal cortex, amygdala, anterior cingulate cortex, nucleus accumbens, hippocampus, and neuron-enriched cell types. Additional tissues (putamen, cerebellum, etc.) were included if data were available."

Issue 6: Missing data on effect sizes, confidence intervals, and sample sizes for forest plots
- Description: Figures 2–5 show forest plots but do not report: (a) Number of studies per SNP; (b) Total sample size (cases + controls); (c) 95% CI ranges for each included study; (d) Study weights; (e) Heterogeneity statistics. While forest plot visualization is useful, tabular supplementary data are required per PRISMA 2020 for verification and meta-analysis reproducibility.
- Location: Figures 2, 3, 4, 5 (pages 6–7, 10)
- Impact on Validity: Readers cannot verify reported effect estimates, assess which studies contributed most weight, or perform sensitivity analyses.
- Required Author Action:
  - Create Supplementary Tables with columns: Author (year), Population, N_cases, N_controls, Genotype/MAF, Effect measure (OR/RR), 95% CI, P-value, Weight (%) in meta-analysis, Study design, Quality (NOS score)
  - Ensure forest plot axes and legends clearly label: Effect measure (OR), null value (OR=1.0), CI ranges, study names, and subgroup labels
  - Cite table in figure caption: "Forest plot of meta-analysis results; detailed study-level data and heterogeneity statistics provided in Supplementary Table X."

Issue 7: QTL results reporting inconsistency—nominal vs. FDR-significant unclear in text
- Description: QTL results are reported with nominal P-values first, then FDR correction mentioned in parentheses, creating ambiguity. Example: "Rs10761482 demonstrated a notable association with ANK3 expression in both neuron-enriched and neocortical cell populations. In neurons, the variant was associated with increased expression (β = 0.12, SE = 0.06, P = 0.038), although the signal did not remain significant following false discovery rate (FDR) correction (FDR = 0.999)." Readers may misinterpret P=0.038 as significant without reading the parenthetical FDR note. Additionally, FDR=0.999 is extremely conservative and raises questions about the number of tests or the correction method.
- Location: Results, pages 8–11, QTL sections; Tables 2–4
- Impact on Validity: Misinterpretation of statistical significance; readers may cite nominal P-values without recognizing FDR inflation. The extremely high FDR values (0.999, 0.965) suggest liberal multiple testing or lenient FDR control, which should be discussed transparently.
- Required Author Action:
  - In Results and Tables 2–4, clearly label findings: "rs10761482 was nominally significant for ANK3 expression in neocortex (P=0.001) but did not meet FDR-adjusted significance threshold (FDR=0.162, n_tissues_tested=XX). In neurons, the association was nominally significant (P=0.038, FDR=0.999), indicating marginal evidence requiring replication."
  - Explain high FDR values: "The high FDR-adjusted P-values reflect correction for testing across [XX] tissues and cell types within the eQTL modality. Although nominally significant, these associations should be interpreted as suggestive pending validation in independent cohorts."
  - Consider reporting Bonferroni-corrected P-values as a comparison threshold, or provide both for transparency

 Minor Issues & Editorial Suggestions

1. Study characteristics table incomplete: Results reference "Extended dataset 2 (https://doi.org/10.6084/m9.figshare.30090913)" containing study characteristics, but recommend including a summary table in the main Results showing: Author (year), Sample size, Population, Psychiatric disorder, Diagnostic criteria, Genotyping platform, Key findings. This aids reader comprehension without requiring external data access.

2. SNP localization table (Table 1) clarity: Table 1 shows exon/intron positions but does not clearly indicate which SNPs are intronic (regulatory) vs. exonic (potentially protein-coding). Recommend adding column: "Location Type: Intronic (regulatory) / Exonic (coding region)." Currently, all appear intronic per the Results text, but the table formatting is ambiguous.

3. Figure 6 gene map resolution: Figure 6 shows the ANK3 gene structure and SNP positions, but lacks a scale bar and precise bp coordinates on axes. For readers to assess SNP clustering, provide: "Map scale: 1 cm = [X] bp" or "Genomic coordinates (GRCh38.p14, NC_000010.11) on x-axis showing ANK3 locus (60.0–60.7 Mb)." The current visualization does not permit readers to discern SNP clustering in the two introns mentioned in the Discussion.

4. Extended Datasets accessibility: Results reference six Extended datasets hosted on Figshare; verify that all DOI links are active and data are publicly accessible (not under embargo). Recommend also depositing on an institutional repository (e.g., University of Southampton Research Portal) with redundant backup.

5. Missing prediction intervals: Random-effects meta-analyses should report 95% prediction intervals (PI) to characterize the range of effects expected in future populations. PRISMA 2020 item 20 recommends PI reporting. Add to Results tables.

DISCUSSION
Summary Assessment
The Discussion provides thoa ughtful interpretation of meta-analysis findings, appropriately contextualizes population-specific heterogeneity, and offers plausible biological mechanisms for intronic SNP effects. However, the Discussion overstates findings regarding mechanistic causality, does not adequately address limitations (only briefly mentioned), and does not discuss implications for clinical practice or future research directions. Limitations acknowledgment is essential for balanced interpretation.

 Major Issues

Issue 1: Mechanistic claims overstate functional evidence available from QTL data
- Description: Discussion states: "These findings suggest that the risk variants may influence alternative splicing decisions that determine exon 37 inclusion. Such splice isoform shifts could lead to changes in the intracellular distribution of Ankyrin-G, potentially altering its ability to cluster ion channels at the AIS and thereby affecting neuronal polarity, excitability, or signaling, leading to cellular and circuit-level changes in neuronal signalling, and to disease in the long term." While plausible, this mechanism is speculative. QTL data show association between SNP genotype and splicing patterns (sQTL), not causality. The Discussion does not acknowledge: (a) sQTL associations are correlative; (b) splicing QTL may tag regulatory variants in linkage disequilibrium (not the SNPs themselves); (c) experimental validation (CRISPR perturbation, reporter assays) is required to establish causality.
- Location: Discussion, page 12
- Impact on Validity: Readers may over-interpret QTL associations as causal proof of mechanism, leading to unwarranted biological claims or premature drug development investment targeting these variants. This is especially problematic for intronic variants, which are rarely directly causal but instead mark regulatory regions.
- Required Author Action:
  - Temper mechanistic claims: "These findings provide suggestive evidence that the identified SNPs may regulate ANK3 splicing in a brain region–specific manner. The sQTL signal at rs28932171, rs11599164, and rs17208576 in cerebellum suggests that these intronic variants may affect alternative exon 37 inclusion, potentially altering isoform distribution. However, these are correlative associations; functional validation through CRISPR perturbation, reporter assays, or allele-specific expression studies in neural cells would be required to confirm mechanistic causality. Additionally, the identified SNPs may not be the causal variants but rather proxy markers in linkage disequilibrium with unmapped regulatory elements."
  - Clarify mechanistic pathway: "Although we cannot infer causality from QTL associations alone, the biological plausibility of our findings is supported by: (1) the documented role of exon 37 in determining AIS-targeted vs. postsynaptic ANK3 isoforms; (2) the functional relevance of AIS ion channel clustering to neuronal excitability; (3) the involvement of AIS dysfunction in neuropsychiatric phenotypes (autism spectrum disorder, epilepsy, ataxia)."

Issue 2: The limitations section is inadequate; key methodological limitations are not discussed
- Description: Discussion briefly states: "Nonetheless, our study has limitations. We did not perform genotype-stratified analyses, which could refine population-specific effects, nor did we incorporate epigenetic QTLs—such as methylation QTLs—that may modulate ANK3 expression." However, several major limitations are not mentioned:
  - (a) Meta-analysis sample sizes are small for some SNP-disorder combinations (e.g., PTSD variants from single studies), limiting statistical power and generalizability
  - (b) Psychiatric phenotypes may differ across studies in diagnostic criteria, severity, age of onset, subtypes (bipolar I vs. II, schizophrenia subtypes), reducing phenotype homogeneity
  - (c) QTL samples are predominantly of European ancestry; regulatory effects may not generalize to non-European populations
  - (d) No causal inference framework employed (e.g., Mendelian randomization); associations may be confounded by linkage disequilibrium or population stratification
  - (e) Candidate SNP selection was driven by meta-analysis findings, potentially limiting the discovery of additional regulatory variants
  - (f) Publication bias was not formally assessed due to the small sample size of the meta-analyses
- Location: Discussion, pages 12–13
- Impact on Validity: Overstated generalizability; readers may not appreciate limitations of effect estimates or applicability to diverse populations. Transparent limitation discussion is essential for balanced interpretation per PRISMA 2020 item 24.
- Required Author Action:
  - Expand Limitations section to include:
Our study has several limitations that warrant acknowledgment. First, meta-analysis sample sizes vary considerably across SNP-disorder combinations, from single studies (PTSD variants) to nine studies (rs10994336 in bipolar disorder), limiting statistical power for smaller subgroups and precision of effect estimates. Second, psychiatric phenotypes differed across studies in diagnostic criteria (DSM versions), assessment methods (clinical interview vs. medical record), and sample composition (age, sex, severity), introducing phenotypic heterogeneity that may attenuate effect estimates. Third, the majority of included studies were conducted in European or East Asian populations; regulatory QTL effects (derived predominantly from GTEx v10, which has limited ancestry diversity) may not generalize to African, Latin American, or South Asian populations, limiting the diversity of our sample and the applicability of findings to underrepresented populations. Fourth, we did not employ causal inference methods (e.g., Mendelian randomization, Bayesian colocalization) to distinguish between true causal variants and proxy markers in linkage disequilibrium, limiting mechanistic inference. Fifth, publication bias was not formally assessed using standard methods due to small sample meta-analyses (n<10 studies per SNP-disorder pair); small-study effects or selective reporting of significant findings may bias effect estimates upward. Sixth, the selection of ten SNPs for QTL analysis was based on meta-analysis findings, potentially limiting the discovery of additional regulatory variants not tested in sufficient studies to meet meta-analysis criteria.

Issue 3: Clinical implications and future research directions were not discussed
- Description: Discussion does not address: (a) How these findings might inform psychiatric risk assessment or precision medicine; (b) Whether identified variants might be therapeutic targets; (c) What follow-up experiments are needed to validate findings; (d) Whether findings have implications for psychiatric nosology or endophenotype definition. PRISMA 2020 and editorial best practices recommend discussing clinical/research implications.
- Location: Discussion, pages 12–13
- Impact on Validity: Readers cannot assess practical significance or next steps for the field. Leaves open questions about translational relevance.
- Required Author Action:
  - Add paragraph on Clinical/Research Implications: "Clinical implications of these findings remain limited at present. While ANK3 variants are associated with psychiatric disorder risk in the populations studied, effect sizes are modest (ORs~1.1–1.3), and no individual genetic variants are sufficiently predictive to warrant clinical genetic testing in current practice. However, these findings contribute to the polygenic risk architecture of psychiatric disorders and may inform larger risk score models. From a research perspective, the identification of tissue-specific regulatory effects (particularly the robust cerebellar sQTL signals for PTSD variants) suggests that ANK3-mediated splicing alterations in specific brain regions may contribute to psychiatric vulnerability. Future studies should: (1) Validate sQTL and eQTL findings in independent neural cell models (iPSC-derived neurons, primary neural cultures) using CRISPR perturbation or reporter assays; (2) Assess whether ANK3 isoform shifts (particularly affecting AIS-localized vs. postsynaptic forms) are detectable in postmortem psychiatric tissue or neuroimaging-derived neural populations; (3) Determine whether ANK3 expression or isoform usage correlates with psychiatric symptoms or cognitive endophenotypes; (4) Conduct cross-ancestry validation in African, Latin American, and South Asian cohorts to assess effect heterogeneity and generalizability."

Issue 4: Prior ANK3 meta-analyses were not comprehensively compared
- Description: Discussion notes: "Although there is no previous comprehensive meta-analysis and systematic review specifically addressing the relationship between ANK3 variants and multiple psychiatric disorders, prior studies have performed meta-analyses focused on individual SNPs—such as rs10994336 in bipolar disorder." However, no detailed comparison of current findings with the prior rs10994336 meta-analysis (Roby, 2017) is provided. For example, do effect estimates differ, and if so, why? Are confidence intervals narrower (due to inclusion of additional studies)? This comparison is important for assessing consistency and meta-meta-analysis trends.
- Location: Discussion, page 13
- Impact on Validity: Readers cannot assess whether findings are novel or consistent with prior work, which affects the interpretation of the contribution's significance.
- Required Author Action:
  - Compare current rs10994336 meta-analysis (OR=1.28, 95% CI: 1.13–1.45, P=1.04×10⁻⁴, n=9 studies) with Roby (2017): "The prior meta-analysis by Roby (2017) reported rs10994336 in bipolar disorder with effect estimate [cite results]. Our expanded meta-analysis, including [X additional studies since 2017], yields a slightly [larger/smaller] effect estimate (OR 1.28 vs. 1. XX in Roby), with [narrower/wider] confidence intervals, reflecting increased sample size and improved precision. The [consistency/divergence] of these findings suggests [stable effect / population-specific variation]."

Issue 5: The population stratification discussion lacks clarity on genetic background mechanisms
- Description: Discussion addresses population-specific heterogeneity (e.g., rs10761482 showing opposite effects in Chinese vs. Iranian populations) and mentions "different populations and genetic backgrounds," but does not clearly explain mechanisms for the opposite direction of effects. This could reflect: (a) genuine allelic heterogeneity (different causal variants in LD with the index SNP in different populations); (b) population-specific linkage disequilibrium structure; (c) gene-by-environment interactions (e.g., different environmental prevalences of risk factors); (d) publication bias or selection bias in study sampling. Discussion is vague and would benefit from explicit hypothesis testing or LD mapping.
- Location: Discussion, pages 12–13
- Impact on Validity: Readers may attribute heterogeneity to vague "population differences" without understanding genetic mechanisms, limiting insight into the true sources of discrepancy.
- Required Author Action:
  - Expand population stratification discussion: "The opposite direction of effect for rs10761482 in Chinese vs. Iranian schizophrenia cohorts (protective in Chinese: OR=0.74; risk-increasing in Iranian: OR=1.57) could reflect several mechanisms: First, linkage disequilibrium structure varies across populations; rs10761482 may be in LD with different causal variants in East Asian vs. West Asian/Middle Eastern haplotypes, leading to apparent direction reversal despite association with the same causal locus. Second, population-specific allele frequencies may alter statistical power and effect estimate precision; rare variants in one population may be moderately common in another. Third, true population-specific effects could arise from gene-by-environment interactions; for example, if environmental stressors associated with schizophrenia risk differ between Chinese and Iranian populations, ANK3 variants may modulate risk differentially depending on environmental context. To distinguish between these mechanisms, future studies could: (1) perform fine-mapping and regional LD analysis in both populations to identify potential causal variants; (2) assess replication in independent samples separately in Chinese and Iranian cohorts; (3) investigate environmental risk factor distributions and potential interaction with ANK3 genotypes."

 Minor Issues & Editorial Suggestions

1. Biological plausibility framing: Discussion appropriately contextualizes the role of the neocortex and cerebellum in psychiatric disorders; however, consider strengthening with recent literature. For example, cerebellar involvement in psychiatric disorders (schizophrenia, autism, depression) is increasingly recognized; cite recent reviews such as Schmahmann JD. Proc Natl Acad Sci USA. 2019;116(19):9375–9384 (cerebellar contribution to psychiatric disorders).

2. Comparison with meQTL findings: Discussion mentions "a previous methylation QTL (meQTL) study, which reported that rs10994336 was significantly associated with ANK3 methylation levels in the prefrontal cortex." Expanding this comparison would strengthen functional annotation: "Integrating our findings with prior meQTL results suggests multi-layered regulatory control of ANK3: methylation-mediated transcriptional regulation (meQTL) and post-transcriptional regulation (eQTL, sQTL). This multi-layered regulation may reflect the critical role of ANK3 in neuronal development and circuit formation, processes requiring precise dosage control."

3. Intronic SNP concentration discussion: Discussion notes "SNPs in both introns could mean that these regions are evolutionarily conserved and therefore may have some functionality." This is reasonable, but could be strengthened by referencing conservation metrics (PhyloCSF, GERP, etc.) or checking whether intronic sequences show evidence of constraint in gnomAD or ClinVar databases.

4. Missing discussion of psychiatric nosology implications: Given that ANK3 variants are associated with multiple psychiatric disorders (bipolar, schizophrenia, depression, PTSD), discuss whether this supports a "transdiagnostic" genetic architecture: "The association of ANK3 variants with multiple psychiatric conditions raises the possibility that ANK3-mediated neuronal dysfunction contributes to shared biological pathways underlying diverse psychiatric phenotypes, supporting a transdiagnostic genetic model of psychiatric illness. This is consistent with recent evidence that psychiatric disorders share common genetic risk factors (Lee et al., 2013; Cross-Disorder Group of the Psychiatric Genomics Consortium)."

CONCLUSIONS

 Summary Assessment

The Conclusion succinctly summarizes key findings (ANK3 intronic variants associated with psychiatric disorders, tissue-specific regulatory effects) and notes the functional relevance of QTL integration. However, the Conclusion is brief and lacks concrete next steps or implications. Recommendations for future research are absent, and the scope of generalizability is not stated.

 Major Issues

Issue 1: The conclusion does not specify the scope of evidence or generalizability
- Description: Conclusion states: "Multiple intronic ANK3 SNPs are significantly associated with psychiatric disorders and display tissue-specific regulatory effects." This statement combines FDR-significant findings (sQTL in cerebellum) with nominal-only findings (eQTL, apaQTL) without distinction. Readers unfamiliar with statistical thresholds may assume all findings are robustly supported; however, effect sizes vary, and some are not validated across populations.
- Location: Conclusion, page 14
- Impact on Validity: Over-interpretation of evidence; readers may attribute equal confidence to FDR-significant sQTL results and nominal eQTL findings.
- Required Author Action:
  - Revise to specify evidence tiers: "Multiple intronic ANK3 SNPs are associated with psychiatric disorders: bipolar disorder and schizophrenia meta-analyses included up to nine studies for individual SNPs, while PTSD variants are based on single or small-sample analyses requiring replication. ANK3 variants display tissue-specific regulatory effects, with the most robust evidence for cerebellar splicing effects (FDR=0.010), modest evidence for neocortical expression effects (FDR=0.162), and nominal evidence for alternative polyadenylation effects (FDR=0.965) that require independent validation."

Issue 2: Conclusion lacks actionable recommendations or research directions
- Description: Conclusion ends: "Integrating QTL data provides insights into their potential functional roles in neuropsychiatric pathogenesis." This is a very general statement offering no specific next steps, no recommendations for clinical translation, no discussion of required validation studies, and no timeline for follow-up research.
- Location: Conclusion, page 14
- Impact on Validity: Readers do not understand what future work is needed to advance the field or translate findings into clinical practice.
- Required Author Action:
  - Expand Conclusion with actionable recommendations:
Integrating QTL data with genetic association findings provides insights into the tissue-specific regulatory mechanisms by which ANK3 genetic variants may contribute to psychiatric disorder risk. However, several steps are necessary to translate these findings toward clinical or therapeutic application: (1) Experimental validation of sQTL and eQTL associations using CRISPR perturbation in primary neural cells or iPSC-derived neurons to establish causality; (2) Cross-ancestry replication of meta-analysis findings in non-European populations to assess effect heterogeneity and generalizability; (3) Integration of ANK3 variants into polygenic risk score models for psychiatric disorders to assess predictive accuracy; (4) Investigation of potential gene-by-environment or gene-by-sex interactions that may modify ANK3 variant effects; (5) Functional characterization of ANK3 isoform distributions in postmortem psychiatric tissue to link genetic associations with pathological phenotypes. Until such validation is complete, ANK3 variants should be considered part of the polygenic architecture underlying psychiatric disease risk rather than individual targets for clinical stratification or drug development.

 Minor Issues & Editorial Suggestions

1. Plural vs. singular reference: Conclusion says "Multiple intronic ANK3 SNPs," but Results identify 10 SNPs total. Consider specifying: "At least 10 intronic ANK3 variants, identified across four psychiatric disorders, display significant genetic associations..."

2. "Insights" vagueness: The phrase "provides insights into...functional roles" is passive and vague. Consider: "elucidate potential mechanisms by which ANK3 genetic variants contribute to psychiatric disease" or "enable hypothesis generation regarding ANK3-mediated neuronal dysfunction in psychiatric pathogenesis."

FIGURES & TABLES
Summary Assessment
Figures 1–7 are generally clear and appropriately support the manuscript text; however, resolution, labeling, and data transparency issues reduce their utility. Tables 1–4 are well-formatted but lack supplementary detail (cell counts, weights, heterogeneity statistics) required for full reproducibility and meta-analysis verification.
Major Issues
Issue 1: Forest plots (Figures 2–5) lack essential statistical annotations
- Description: Figures 2–5 show forest plots with individual study effect estimates and pooled OR, but lack: (a) heterogeneity statistics (I², τ²) on the plot or in caption; (b) study weights (proportional to bubble size); (c) number of studies and combined sample size; (d) confidence intervals for individual studies (confidence intervals should be shown on plot); (e) funnel plot estimates or publication bias commentary.
- Location: Figures 2, 3, 4, 5 (pages 6–7, 10)
- Impact on Validity: Readers cannot assess statistical rigor or identify studies with outlier effects. Standard forest plot reporting per Cochrane/PRISMA requires confidence intervals and heterogeneity statistics.
- Required Author Action:
  - Add to each forest plot caption: "Heterogeneity statistics: I² = XX%, τ² = XX, Q-test p-value = X.XXX. Pooled effect estimated using random-effects meta-analysis. Individual study squares represent point estimates (OR); horizontal lines span 95% confidence intervals; the diamond represents the pooled estimate and 95% CI. Study labels show author (year); sample sizes (cases/controls) are indicated in parentheses."
  - Provide forest plot data tables in Supplementary Materials showing: Study, N_cases, N_controls, OR, 95% CI, Weight (%)
  - If publication bias is suspected, add funnel plot and Egger test results to supplement forest plots

Issue 2: Figure 1 (literature screening flow diagram) is missing data
- Description: Already detailed in the Results section, Issue 2 above. Figure 1 is overly simplified and lacks exclusion reason tallies.
- Location: Figure 1, page 5
- Impact on Validity: [See Results Issue 2 for full discussion and required action]
- Required Author Action: [As specified in Results Issue 2]

Issue 3: Figure 6 (SNP localization on ANK3 gene) lacks scale and clarity
- Description: Figure 6 displays the ANK3 gene structure with SNP positions; however, the figure lacks: (a) a genomic scale bar (bp or Mb); (b) axis labels with coordinates (GRCh38.p14); (c) exon numbering clarity; (d) visual distinction between major and minor SNP clusters. Currently, SNP positions are not easily readable without high-resolution magnification.
- Location: Figure 6, page 8
- Impact on Validity: Readers cannot discern the scale of SNP clustering or the gene's size relative to SNPs. The Discussion emphasizes "concentration of SNPs in both introns," but Figure 6 does not make this apparent.
- Required Author Action:
  - Redraw Figure 6 with: (1) Genomic coordinates on x-axis (e.g., "Chr10: 60.2 Mb – 60.7 Mb"); (2) Scale bar (e.g., "1 Mb"); (3) Exon numbering above gene structure; (4) Intron numbers or names; (5) SNP labels with rsID and associated disorder color-coding (e.g., red for bipolar, blue for schizophrenia, green for PTSD); (6) clear visual distinction between the two main introns harboring clusters (Intron 2–3 and Intron 36–37)
  - High-resolution version suitable for publication (at least 300 dpi)

Issue 4: Table 1 (SNP localization summary) lacks phenotype specificity
- Description: Table 1 summarizes SNP positions in exons/introns and associates them with psychiatric disorder columns (Bipolar disorder, Schizophrenia, Major depression, Posttraumatic stress disorder). However, the table does not indicate which SNPs are significant in each disorder (italics or bold formatting would help), and the distinction between meta-analysis significant and single-study reported variants is not clear.
- Location: Table 1, page 9
- Impact on Validity: Readers cannot quickly identify which SNPs are robustly associated with which disorders. Table readability is reduced.
- Required Author Action:
  - Reformat Table 1 as follows:
    ```
    | SNP | Location | Chr:bp (GRCh38) | Bipolar Disorder | Schizophrenia | Major Depression | PTSD |
    |--|-|--||-|||
    | rs10994336 | Intron 2–3 | 10:60,XXX | OR=1.28 (n=9) | ns (n=X) | ns (n=X) | — |
    | rs9804190 | Intron 36–37 | 10:60,XXX | OR=0.88 (Eur) (n=3) | ns | — | nominal |
    | rs10994359 | Intron 2–3 | 10:60,XXX | — | — | OR=0.69 (Asian) (n=X) | — |
   
    Significance levels:  FDR<0.001;  FDR<0.01;  FDR<0.05; ns, not significant in meta-analysis; —, not tested. Abbreviations: OR, odds ratio; Eur, European populations; n, number of independent studies in meta-analysis.
- Add footnote clarifying which results are from meta-analysis (n≥2 studies) vs. synthetic review (n=1 study)

Issue 5: Tables 2–4 (QTL results) do not show sufficient context (number of tissues tested, n values)
- Description: Tables 2–4 report QTL effect estimates (β, SE, P-value, FDR) for individual SNP-tissue pairs, but lack context: How many tissues were tested in total for eQTL? How many SNPs were tested? Tables appear to show only significant or selected findings; complete results are referenced to the Extended datasets, but summary tables in the main text would improve interpretation.
- Location: Tables 2, 3, 4 (pages 10–11)
- Impact on Validity: Readers cannot assess the total multiple testing burden or whether selective reporting of results has occurred. For example, if 10 SNPs × 50 tissues = 500 eQTL tests were performed, FDR correction should reflect this, and the single significant eQTL for rs10761482 should be interpreted cautiously.
- Required Author Action:
  - Add table header or caption: "eQTL analysis: 10 SNPs tested across 47 brain tissues/cell types in the eQTL Catalogue = 470 SNP-tissue pairs tested. FDR correction applied within eQTL modality; FDR<0.05 considered significant. Results shown for SNPs with nominal P<0.05; complete results in Extended dataset 4."
  - Create supplementary summary table showing: (1) Total number of SNP-tissue pairs tested per QTL modality; (2) Number of significant associations (FDR<0.05); (3) Number of nominally significant but FDR-nonsignificant findings; (4) Range of effect sizes observed
  - Provide complete results tables for eQTL, apaQTL, sQTL in Supplementary Materials or Extended datasets

 Minor Issues & Editorial Suggestions

1. Figure legibility: Ensure all figure text (axis labels, legends, SNP labels) is legible at publication size (typically 8–10pt font for figures). Current Figures 2–5 forest plot study labels may be too small.

2. Color coding consistency: If different colors are used for different disorders across figures (e.g., red for bipolar, blue for schizophrenia in Figure 6), ensure the color scheme is consistent and colorblind-friendly (avoid red-green combinations).

3. Figure panel labels: Ensure all multi-panel figures (e.g., Figure 7 with panels A–D, Figure 2 with panels A–D) are clearly labeled with capital letters and referenced in the caption.

4. Supplementary materials reference: Ensure all references to "Extended dataset" include active DOI links and verify Figshare accessibility at submission.


ETHICAL CONSIDERATIONS, FUNDING, DATA AVAILABILITY

Summary Assessment

The manuscript appropriately states that the study used only publicly available summary-level data and did not involve human participants, no ethics approval was required. Funding and competing interests disclosures are provided. However, the data availability statement could be more explicit regarding which raw data are available vs. which are restricted, and the PROSPERO registration status is not mentioned.

Major Issues

Issue 1: PROSPERO registration status not reported
- Description: The manuscript does not state whether this systematic review was registered with PROSPERO (International Prospective Register of Systematic Reviews) before publication. PRISMA 2020 item 4 requires reporting of PROSPERO registration number and registration date. Pre-registration is essential for reducing bias and demonstrating that analyses were not post-hoc selected based on observed data.
- Location: Methods section, or missing entirely
- Impact on Validity: Inability to verify whether eligibility criteria, quality assessment, and statistical analysis plans were pre-registered vs. decided after observing data. Increases risk of selective reporting.
- Required Author Action:
  - If PROSPERO registration was completed: Include PROSPERO registration number (e.g., "PROSPERO registration: CRD42024XXXXXXX") in Methods and cite in Results (e.g., "This systematic review was registered with PROSPERO before the start of study selection (PROSPERO registration: CRDXXXXXXX, registered [date]).")
  - If PROSPERO registration was not completed (post-hoc registration of completed review is possible): State in Methods: "This systematic review was completed before PROSPERO registration but has been registered retroactively (PROSPERO registration: CRDXXXXXXX). The predetermined protocol [if available as supplementary document] specified the eligibility criteria, search strategy, quality assessment tool (Newcastle-Ottawa Scale), and planned analyses."
  - If PROSPERO registration was not completed and the pre-registration document is not available: Acknowledge as a limitation and state that future reviews will be pre-registered

Issue 2: Data availability statement incomplete regarding access restrictions
- Description: Data Availability section states: "The data used in this meta-analysis and QTLs analysis, including the eQTL Catalogue and GTEx v10, were obtained from published articles and public databases... All datasets are released under the terms of the Creative Commons Attribution 4.0 International licence (CC BY 4.0)." However, it does not specify: (a) Are the extracted datasets (study-level ORs, 95% CIs) provided as supplementary files or only on Figshare? (b) Can the raw individual-participant data from included studies be accessed, and if not, why? (c) Are the R scripts and data parsing code available for reproducibility? (d) Are there any data use restrictions from the eQTL Catalogue or GTEx v10?
- Location: Data Availability section, pages 13–14
- Impact on Validity: Readers cannot easily access or verify the extracted data, limiting reproducibility and independent verification of reported findings.
- Required Author Action:
  - Expand Data Availability statement to:
  
    Data Availability:
    This study is based entirely on publicly available summary-level data from published literature and public genomic databases. No new primary data were generated. The extracted meta-analysis datasets, including study-level effect estimates (odds ratios, 95% confidence intervals, P-values) for all included studies and SNP-disorder combinations, are openly available on Figshare under Creative Commons Attribution 4.0 International (CC BY 4.0) license:
   
    - Extended dataset 1 (Meta-analysis extracted data): https://doi.org/10.6084/m9.figshare.29886620
    - Extended dataset 2 (Study characteristics): https://doi.org/10.6084/m9.figshare.30090913
    - Extended dataset 3 (Meta-analysis results summary): https://doi.org/10.6084/m9.figshare.30090964
    - Extended dataset 4 (eQTL analysis results): https://doi.org/10.6084/m9.figshare.29901494
    - Extended dataset 5 (apaQTL analysis results): https://doi.org/10.6084/m9.figshare.29901641
    - Extended dataset 6 (sQTL analysis results): https://doi.org/10.6084/m9.figshare.29901662
   
    All R and Python scripts used for data parsing, harmonization, and QTL analysis are available on GitHub at [repository URL] and archived on Figshare [DOI] under a CC BY 4.0 license.
   
    The eQTL Catalogue and GTEx v10 summary statistics are publicly available through their respective portals (https://www.ebi.ac.uk/eqtl/ and https://gtexportal.org/) under their respective data use agreements. No restrictions apply to their use for secondary analysis and research purposes.
   
    Individual-level genotype and phenotype data from the included genome-wide association studies are typically not publicly available due to privacy regulations (e.g., GDPR); however, summary-level statistics have been extracted and are provided in our datasets. Access to individual-level GWAS data may be available through the original study authors or through the Psychiatric Genomics Consortium (https://www.med.unc.edu/pgc/) with appropriate data use agreements.
   
    Computed outputs (forest plots, QTL heatmaps, supplementary tables) are provided as Supplementary Materials and Extended datasets.

Issue 3: Funding statement adequacy
- Description: Grant information states: "The author(s) declared that no grants were involved in supporting this work." However, the acknowledgments section mentions: "TG gratefully acknowledges the invaluable guidance and support provided by his supervisors and research team throughout the course of his PhD studies. TG also extends his sincere appreciation to the China Scholarship Council for their financial support." This is contradictory: if the China Scholarship Council provided financial support, it should be listed under funding, not only in acknowledgments.
- Location: Grant information section (page 2) and Acknowledgments (page 15)
- Impact on Validity: Incomplete funding disclosure; readers may not recognize the source of financial support, affecting the assessment of potential bias. ICMJE guidelines require complete disclosure of all funding sources.
- Required Author Action:
  - Revise Grant information to: "Grant information: TG was supported by the China Scholarship Council (CSC) PhD Scholarship (2019–2023). YW acknowledges support from [list any research grants or institutional funding]. KD acknowledges support from [list any research grants or institutional funding]. No additional grants were involved in supporting this work."
  - Streamline Acknowledgments to: "TG thanks his supervisors for mentorship and research team guidance. Figshare hosting and extended data management were supported by [institutional funding or F1000Research open access initiative]."

 Minor Issues & Editorial Suggestions

1. Open Peer Review Option: F1000Research is known for open peer review, where reviewer reports are published with the article. The manuscript does not mention whether the authors have opted into open peer review. Consider: "Authors consent to open peer review; reviewer reports will be published alongside this article" (or alternative statement from F1000Research policy).

2. Ethics approval statement: Current ethics statement appropriately notes: "This study was based entirely on publicly available summary-level genomic data... Therefore, no ethical approval or informed consent was required." Consider adding a reference to the ICMJE guidelines: "In accordance with ICMJE guidelines (http://www.icmje.org/recommendations/browse/roles-responsibilities/author-responsibilities-conflicts-of-interest.html), this meta-analysis did not require ethics approval as no primary data were collected."
LINE-BY-LINE EDITORIAL SUGGESTIONS

 Abstract

 Current Text (Methods Section):
"A meta-analysis and systematic literature review were conducted to identify disease-associated SNPs in ANK3. QTL analyses were performed using datasets from the eQTL Catalogue and GTEx v10 to evaluate regulatory effects in brain tissues."

 Suggested Replacement:
"A systematic literature review and meta-analysis were conducted to identify ANK3 single-nucleotide polymorphisms (SNPs) significantly associated with psychiatric disorders across published studies (PubMed, Web of Science, MEDLINE; search period: January 2000–September 2025; n=27 included studies). Population-stratified meta-analyses were performed using odds ratios from included studies and random-effects models. Secondary expression quantitative trait locus (eQTL), alternative polyadenylation QTL (apaQTL), and splicing QTL (sQTL) analyses were conducted using public datasets (eQTL Catalogue, GTEx v10) to characterize tissue-specific regulatory effects of disease-associated variants in the human brain."

 Current Text (Results Section):
"In bipolar disorder, rs10994336 (OR = 1.28, P = 1.04×10−4) was significant in European+North American populations; rs9804190 was protective in Europeans (OR = 0.88, P = 0.020)..."

 Suggested Replacement (with specificity of significance):
"Meta-analysis identified ten ANK3 variants significantly associated with psychiatric disorders. In bipolar disorder, rs10994336 (OR = 1.28, 95% CI: 1.13–1.45, P = 1.04×10⁻⁴) emerged as a significant risk variant in European and North American populations (n=9 studies); this association was not replicated in Asian cohorts (P = 0.600, n=X studies). Rs9804190 showed significant protective effects in European studies (OR = 0.88, 95% CI: 0.79–0.98, P = 0.020, n=3 studies); however, this association did not extend to combined European and North American analysis (P = 0.173). Rs1938526 demonstrated robust cross-population associations in both European/North American (OR = 1.32, 95% CI: 1.22–1.43, P = 7.26×10⁻¹²) and Asian populations (OR = 1.08, 95% CI: 1.04–1.12, P = 4.68×10⁻¹²), with pooled effect estimate OR = 1.17 (95% CI: 1.05–1.29, P = 0.003)..."


 Introduction

 Current Text:
"Over the past decade, genome-wide association studies have identified significant associations between ANK3 single nucleotide polymorphisms (SNPs) and a range of psychiatric disorders, including bipolar disorder (BD), schizophrenia (SZ), major depressive disorder (MD), and post-traumatic stress disorder (PTSD). These findings suggest that genetic variation within the ANK3 locus may contribute to shared pathophysiological mechanisms underlying multiple neuropsychiatric phenotypes. However, despite accumulating evidence, considerable inconsistencies remain regarding which ANK3 SNPs are most robustly associated with psychiatric disorders and whether these associations are consistent across diverse populations."

 Suggested Replacement (with specific examples):
"Over the past decade, genome-wide association studies (GWAS) have identified multiple significant associations between ANK3 single nucleotide polymorphisms (SNPs) and a range of psychiatric disorders, including bipolar disorder (BD), schizophrenia (SZ), major depressive disorder (MD), and post-traumatic stress disorder (PTSD). These findings suggest that genetic variation within the ANK3 locus may contribute to shared pathophysiological mechanisms underlying multiple neuropsychiatric phenotypes. However, considerable inconsistencies remain in the literature: for example, rs10994336 was significantly associated with bipolar disorder in European cohorts (Ferreira et al., 2008; Scott et al., 2009) but not in East Asian populations (Takata et al., 2011; Lee et al., 2011), and rs10761482 showed opposite associations in Chinese (protective) versus Iranian (risk-increasing) schizophrenia cohorts (Dashtban et al., 2022; Karimian et al., 2020). Such population-specific effect heterogeneity suggests that ANK3 genetic architecture may vary across ancestries, highlighting the need for systematic synthesis of findings with explicit population stratification and comprehensive functional characterization of disease-associated variants."



 Methods

 Current Text (Search Strategy):
"In accordance with the predefined inclusion criteria, two independent researchers (TG and YW) systematically searched PubMed, Web of Science, and MEDLINE using combinations of the terms 'genome-wide association study' with 'ANK3', as well as 'nucleotide polymorphisms' with 'ankyrin 3' or 'Ankyrin-G'. Relevant studies were screened, and data were extracted independently by both reviewers using a standardized data extraction form."

 Suggested Replacement:
"In accordance with the predefined inclusion criteria and PRISMA 2020 guidelines, two independent researchers (TG and YW) systematically searched PubMed, Web of Science, and MEDLINE from 1 January 2000 through 30 September 2025. The search strategy combined controlled vocabulary and text terms as follows:

PubMed search string:
`(ANK3 OR "ankyrin 3" OR "ankyrin-3" OR "ankyrin G" OR "ankyrin-G" OR ANKRD4 OR "ankyrin 3 gene") AND (polymorphism OR SNP OR genotype OR allele OR "rs[0-9]" OR association OR GWAS) AND (psychiatric OR schizophrenia OR bipolar OR depression OR PTSD OR "post traumatic" OR "mental disorder")`

Web of Science topic search:
`TS=(ANK3 OR "ankyrin 3" OR "ankyrin-G") AND TS=(SNP OR polymorphism OR genotype) AND TS=(schizophrenia OR bipolar OR depression OR psychiatric OR PTSD)`

Searches were limited to peer-reviewed publications in English. Retrieved references were imported into a reference management system (Mendeley or Zotero), duplicates removed, and remaining articles (n=1,140) underwent two-stage screening: (1) title and abstract review by both researchers independently, and (2) full-text assessment of potentially eligible articles. Disagreements were resolved through discussion with a third researcher (KD) until consensus was achieved. A Preferred Reporting Items for Systematic Review and Meta-Analysis (PRISMA) 2020 flow diagram documenting the study selection process is provided in Figure 1."



 Results

 Current Text (Study Selection):
"Using the predefined search strategy, a total of 1,140 articles were retrieved from the selected databases. Following title and abstract screening, duplicate entries were removed. Full-text review resulted in the exclusion of 214 studies, primarily due to methodological insufficiencies or unclear study protocols. Ultimately, 27 English-language studies met the inclusion criteria and were retained for analysis."

 Suggested Replacement:
"Using the predefined search strategy, a total of 1,140 unique articles were retrieved from PubMed (n=XXX), Web of Science (n=XXX), and MEDLINE (n=XXX) [insert actual numbers from original search results]. After removal of XXX duplicate records, 1,XXX articles underwent title and abstract screening, resulting in XXX articles advanced to full-text review. Full-text assessment resulted in exclusion of 214 studies for the following reasons: [n=XX] did not specifically test ANK3 SNP associations; [n=XX] focused on non-psychiatric outcomes; [n=XX] were reviews, case reports, or expert commentaries; [n=XX] lacked clear information on participant geography; [n=XX] reported hazard ratios rather than odds ratios; [n=XX] other (specify). Ultimately, 27 studies met all inclusion criteria and were retained for systematic review and meta-analysis. All included studies achieved quality scores ≥5/9 on the Newcastle-Ottawa Scale, indicating adequate study quality (Supplementary Table X). The study selection process is documented in the PRISMA 2020 flow diagram (Figure 1)."

Discussion

Current Text (Mechanistic Claims):
"Interestingly, three of the SNPs significantly associated with psychiatric disorders—rs28932171, rs11599164, and rs17208576—that are located in the intronic region between exon 36 and exon 37 exhibit strong splicing QTL (sQTL) signals specifically in cerebellum tissue (Table 4). These findings suggest that the risk variants may influence alternative splicing decisions that determine exon 37 inclusion. Such splice isoform shifts could lead to changes in the intracellular distribution of Ankyrin-G, potentially altering its ability to cluster ion channels at the AIS and thereby affecting neuronal polarity, excitability, or signaling, leading to cellular and circuit-level changes in neuronal signalling, and to disease in the long term."

 Suggested Replacement (Tempered Claims with Caution):
"Notably, three ANK3 variants significantly associated with psychiatric disorders—rs28932171, rs11599164, and rs17208576—are located in the intronic region upstream of exon 37 (a brain-specific exon encoding the AIS targeting motif of Ankyrin-G) and show robust evidence of association with alternative splicing in human cerebellum tissue (sQTL FDR=0.010, Table 4). These quantitative trait locus associations provide suggestive evidence that these intronic variants may regulate ANK3 exon 37 inclusion, potentially affecting isoform composition in cerebellum. Given the documented role of exon 37 in determining AIS-localized vs. postsynaptic ANK3 isoforms, and the established importance of AIS ion channel clustering for neuronal excitability, one plausible mechanism is that splice isoform shifts driven by these variants could alter the relative abundance of AIS-targeted and postsynaptic Ankyrin-G pools, thereby modulating neuronal polarity, action potential generation, or circuit connectivity in cerebellum. However, these sQTL associations are correlative rather than causal; the identified SNPs may serve as proxy markers in linkage disequilibrium with unmapped regulatory elements or non-coding RNA genes, rather than being directly functional. Experimental validation through CRISPR perturbation of these intronic elements, allele-specific expression assays, and functional characterization of ANK3 isoform shifts in neural cells would be necessary to establish mechanistic causality. Furthermore, the relevance of cerebellar ANK3 splicing alterations to psychiatric symptomatology remains unknown; future studies directly assessing ANK3 isoform distributions in postmortem psychiatric brain tissue or identifying correlations between ANK3 splicing variants and clinical phenotypes would clarify the functional impact on disease pathogenesis."

Conclusion

 Current Text:
"Multiple intronic ANK3 SNPs are significantly associated with psychiatric disorders and display tissue-specific regulatory effects. Integrating QTL data provides insights into their potential functional roles in neuropsychiatric pathogenesis."

 Suggested Replacement:
"Multiple intronic ANK3 variants demonstrate significant associations with psychiatric disorders, particularly bipolar disorder and schizophrenia, with effect sizes varying across populations. These variants display tissue-specific regulatory effects on ANK3 gene expression and splicing, with the most robust evidence for cerebellar alternative splicing (sQTL FDR=0.010) and modest evidence for neocortical expression regulation (eQTL FDR=0.162). However, several important limitations warrant acknowledgment: (1) meta-analysis sample sizes are modest for some SNP-disorder combinations, limiting statistical power and generalizability, particularly for PTSD variants derived from single studies; (2) psychiatric phenotypes differ across studies in diagnostic criteria and assessment methods, introducing phenotypic heterogeneity; (3) regulatory QTL effects are predominantly derived from European ancestry samples and may not generalize to underrepresented populations. Integrating genetics with multi-layered functional genomics (eQTL, sQTL, apaQTL) provides important insights into potential tissue-specific mechanisms by which ANK3 intronic variants may contribute to psychiatric disorder vulnerability. However, these associations are correlative; future experimental validation in neural cell models, cross-ancestry replication, and investigation of gene-by-environment interactions are necessary to establish causality and clarify the functional role of ANK3 in psychiatric pathogenesis. Integration of ANK3 variants into larger polygenic risk models may improve psychiatric disease prediction, but individual variants are not yet suitable for clinical genetic testing given modest effect sizes. Further research integrating genetics, molecular biology, and psychiatric endophenotyping is essential to translate these findings toward clinical or therapeutic application."

OVERALL RECOMMENDATION

Summary of Key Findings

Strengths:
1. Timely and comprehensive systematic review addressing an important psychiatric genetics question
2. Well-motivated integration of meta-analysis with functional genomics (QTL analyses)
3. Appropriate acknowledgment of population stratification and heterogeneity
4. Transparent reporting of FDR-corrected vs. nominal findings
5. Open data and transparent science practices (Figshare data deposition, CC-BY license)
6. Clear acknowledgment of ethical practices (no human subjects, appropriate data sources)

Critical Weaknesses:
1. Search strategy insufficiently reported: Databases, dates, Boolean operators, MeSH terms, completeness filters not transparent. Risk of selection bias.
2. Heterogeneity statistics omitted from main Results: I² and τ² not reported for meta-analyses, limiting assessment of effect generalizability.
3. PRISMA 2020 non-compliant: PROSPERO registration status not reported; PRISMA checklist not included in submission; explicit PICO statement absent.
4. Mechanistic claims overstated: QTL findings presented as causal evidence; requires tempering with acknowledgment of correlative nature of associations.
5. Limitations inadequately discussed: Small sample sizes (PTSD variants single-study), phenotype heterogeneity, QTL ancestry bias, lack of causal inference, publication bias assessment not conducted.
6. Data transparency gaps: Heterogeneity statistics, individual study data, NOS item-level assessments not in main text; reliance on Extended datasets may limit reproducibility.
7. Conflicting funding disclosures: Grant information states "no grants" while Acknowledgments mention China Scholarship Council support.

Moderate Weaknesses:
1. Clinical implications and future research directions minimally discussed
2. Comparison with prior ANK3 rs10994336 meta-analysis incomplete
3. QTL tissue selection and justification not documented
4. Figures (forest plots, gene map) lack essential statistical annotations and legend detail
5. PTSD findings (three SNPs with identical OR=0.48, P=0.045) require careful interpretation due to small sample sizes

MAJOR REVISIONS REQUIRED

Justification:

The manuscript presents a solid evidence synthesis of ANK3 genetic associations with psychiatric disorders, integrates functional genomics meaningfully, and demonstrates good scientific rigor in several areas (multi-reader screening, quality assessment, QTL multiple-testing correction). However, critical gaps in methodological transparency, compliance with reporting standards (PRISMA 2020, ICMJE), and statistical reporting significantly limit publication readiness. The work is scientifically sound but requires substantial revision for publication in a rigorous, open-access venue such as F1000Research.

Primary Issues Requiring Resolution (Critical):
1. Complete search strategy documentation (databases, dates, Boolean operators, counts per stage)
2. Explicit PICO statement in Introduction or Methods
3. Report heterogeneity statistics (I², τ², Q-test P-values) for all meta-analyses in Results and Tables
4. PROSPERO registration documentation or acknowledgment of post-hoc registration
5. Funding disclosure correction (resolve Grant information vs. Acknowledgments discrepancy)
6. Expanded Limitations section addressing: small sample sizes, phenotype heterogeneity, ancestry bias in QTL, lack of causal inference, publication bias, small-study effects
7. Tempering of mechanistic claims with explicit caveat that QTL associations are correlative, not causal
8. Addition of clinical implications and future research directions discussion

Secondary Issues Requiring Resolution (Major):
1. Comprehensive statistical tables with individual study data, weights, confidence intervals, and heterogeneity statistics for forest plots
2. Forest plots with heterogeneity statistics in captions or overlaid on plots
3. Complete QTL methods detail: tissues/cell types tested, total number of tests, FDR correction rationale, interpretation of nominal vs. FDR-significant findings
4. Risk-of-bias assessment transparency: NOS scores for all 27 studies in supplementary table
5. Clarification of PTSD variant data source and sample size justification

DIRECT QUESTIONS FOR AUTHORS (REVISION RESPONSE)

During revision, authors should address the following numbered questions in a point-by-point response document:

1. PROSPERO Registration: Was this systematic review pre-registered with PROSPERO prior to study selection? If yes, provide registration number (CRD4XXXXXXX). If no, explain why pre-registration was not conducted, and indicate whether post-hoc registration is feasible before publication.

2. Search Strategy Reproducibility: Can authors provide the complete, exact search strings used in PubMed, Web of Science, and MEDLINE, including MeSH terms, field tags, date ranges, and any filters applied (e.g., humans, English language, peer review status)? Have the search strategies been validated against known relevant studies (e.g., Ferreira et al. 2008, Schulze et al. 2009 on ANK3 bipolar disorder)?

3. Heterogeneity Statistics: Authors state that meta-analysis was conducted using the R meta package version 4.18-0. Can authors provide the I² (and 95% CI for I²), τ², Q-test statistic, and Q-test P-value for EACH SNP-disorder meta-analysis (not just those reported in Extended datasets)? Were prediction intervals calculated and reported?

4. HWE Stratification Pre-Specification: In the schizophrenia rs10761482 analysis, was the stratification by Hardy-Weinberg equilibrium (HWE) status predetermined before data extraction, or decided post-hoc after observing HWE violations in Chinese studies? If post-hoc, can authors justify this decision and acknowledge it as an exploratory subgroup analysis?

5. PTSD Sample Sizes: The three PTSD SNPs (rs28932171, rs11599164, rs17208576) are reported with identical OR=0.48 and P=0.045. How many independent studies contributed to each PTSD SNP meta-analysis? Were these derived from a single study or linked studies? If from a single study, this should be clearly labeled as a "synthetic review" finding, not a meta-analysis.

6. QTL Tissue Selection: What was the total number of tissues and cell types tested in the eQTL analysis? For example, was every available tissue in the eQTL Catalogue included, or were specific tissues (e.g., psychiatric-relevant tissues) selected a priori? How many SNP-tissue pairs were tested in total for each QTL modality (eQTL, apaQTL, sQTL)? This number is needed to verify the FDR correction magnitude.

7. Newcastle-Ottawa Scale Scores: Can authors provide a supplementary table with complete NOS itemized scores for all 27 included studies? Specifically, for each study, report scores for: (a) case definition adequacy; (b) case representativeness; (c) control selection; (d) control definition; (e) comparability on design; (f) comparability on analysis; (g) exposure ascertainment; (h) same ascertainment method; (i) non-response rate. Total score (/9) should also be provided.

8. Publication Bias Assessment: Authors state "publication bias was not assessed" because fewer than 10 studies per meta-analysis. However, do funnel plots exist for the larger meta-analyses (e.g., rs10994336 BD with n=9 studies)? Can authors provide visual funnel plots and Egger test results (even with caution regarding small sample limitations) in supplementary materials?

9. Funding Clarification: The Grant information section states "no grants were involved," yet the Acknowledgments state "TG also extends his sincere appreciation to the China Scholarship Council for their financial support." Was the China Scholarship Council PhD scholarship considered "grant funding" by the authors? Please clarify the funding status and revise the Grant information section to accurately reflect all funding sources.

10. Effect Direction Consistency: In schizophrenia rs10761482, Chinese cohorts show a protective effect (OR=0.74) while Iranian cohorts show a risk effect (OR=1.57). Did authors assess whether these could reflect the same causal variant (with differential LD across populations) vs. genuinely opposite effects? Can authors provide linkage disequilibrium data or fine-mapping evidence for this locus across populations?

11. Generalizability of QTL Findings: The eQTL Catalogue and GTEx v10 datasets have limited ancestry diversity, predominantly European. Can authors assess or mention whether eQTL, sQTL, and apaQTL effect directions and magnitudes are likely conserved across ancestry groups? Alternatively, is there published evidence on trans-ethnic generalizability of these specific QTL associations?

12. Causal Inference Frameworks: Did authors consider employing Mendelian randomization or colocalization analyses to distinguish between (a) direct functional variants, (b) regulatory variants affecting ANK3 expression, and (c) proxy markers in linkage disequilibrium? If not, is this a planned direction for future work?