henhaobio
Sign in

bioRxivGeneticsPreprint

Analysis of Rare Coding Variation Identifies New Genetic Contributors to Schizophrenia

Julia SealockiD, Connor DowdiD, Calwing LiaoiD, Robert YeiD, Daniel P. HowriganiD, F. Kyle SatterstromiD, Chiara AuwerxiD, Soyeon KimiD, Cong HuaiiD, Lin He, Melkam Alemayehu KebedeiD, Stella GichuruiD, Rehema M. Mwende, Charles R. J. C. NewtoniD, Nastassja KoeniD, Zukiswa ZingelaiD, Joseph KyebuzibwaiD, Anne StevensoniD, Arsalan HassaniD, Kai WangiD, Makoto AraiiD, Yasue HoriuchiiD, Masanari ItokawaiD, Nora Ayola Serrano, Carrie E. BeardeniD, , Akena Dickens

PDFDOICodeData

In-depth summary

Claude Opus 5.5 · xhigh effort · 7 Nov 2026 · from the full text of version 1 (21 Sept 2026) · CC BY 4.0

Ultra-rare damaging coding variants in 87,959 schizophrenia cases and 150,587 controls implicate 40 genes (16 exome-wide, 24 at 5% FDR), half overlapping autism, developmental delay or bipolar genes.

Common schizophrenia GWAS loci are mostly non-coding with small effects, and the first Schizophrenia Exome Meta-Analysis (SCHEMA 1.0) implicated 10 genes, with follow-up studies adding a few more. SCHEMA 2.0 triples the case count and replaces the MPC-only missense filter with MisMeanRank, a rank average of four pathogenicity scores; case-control data alone now give 16 exome-wide genes versus four from the equivalent SCHEMA 1.0 analysis.

Key results

  1. Shown16 genes pass exome-wide significance and 24 more pass 5% FDR. New exome-wide genes include SCAF1, FYN, PPP3CA, HDAC9, JARID2, ATP9A, PTK2 and SCN2A. CHRM4, the muscarinic receptor gene targeted by the antipsychotic xanomeline/trospium, reaches 5% FDR.

    Bonferroni threshold p < 1.31 x 10^-6; FDR 5% threshold p < 6.84 x 10^-5. Six of the 10 SCHEMA 1.0 exome-wide genes remain exome-wide (SETD1A, HERC1, RB1CC1, SP4, XPO7, CUL1). 35 of 40 genes have OR > 2. A protein-truncating variant (PTV) in any of the 40 genes occurs in 1.5% of cases vs 0.5% of controls, OR 3.43 (95% CI 3.12-3.76). · Abstract; Identifying genes implicated by RVAS; Figure legends

  2. ShownCases carry more ultra-rare damaging variants in constrained genes (pLI ≥ 0.9) than controls, most strongly protein-truncating singletons, with no excess of synonymous singletons. PTV-singleton enrichment is significant in every ancestry group.

    Singletons, cases vs controls: PTV OR = 1.21 (p = 7.92 x 10^-92); PTV + missense OR = 1.19 (p = 5.12 x 10^-154); missense OR = 1.17 (p = 1.75 x 10^-67); synonymous OR = 1.00 (p = 0.982). PTVs with minor allele count (MAC) 2-5: OR = 1.08 (p = 1.67 x 10^-4). Only 555 of 550,916 ultra-rare PTVs (0.10%) exceed 0.1% popmax frequency in gnomAD v4. · Identifying ultra-rare, damaging variants

  3. ShownAdding MisMeanRank-defined damaging missense variants to PTVs strengthens most gene signals, the older MPC filter recovers fewer genes, and adding de novo mutations from trios changes little.

    For 27 of 40 FDR 5% genes the PTV-only signal is weaker than PTV + damaging missense. Using MPC > 3 instead of MisMeanRank ≥ 93 yields 13 exome-wide and 25 FDR 5% genes (vs 16 and 40). Meta-analysing de novo calls from 3,402 SCHEMA 1.0 trios adds only TRIO at exome-wide level (p = 1.07 x 10^-6) and YLPM1 at FDR 5% (p = 6.70 x 10^-5). · Identifying genes implicated by RVAS; Comparing SCHEMA 2.0 to SCHEMA 1.0; Variant annotation

  4. ShownHalf the genes are also implicated in developmental delay/intellectual disability (DD/ID), autism spectrum disorder (ASD) or bipolar disorder; cases carry excess damaging variants in DD/ID and ASD genes. GWAS-locus overlap is at chance, but fine-mapped GWAS genes show excess damaging missense.

    PTV singletons: OR = 1.47 (p = 4.10 x 10^-46) in ASD genes, OR = 1.27 (p = 1.23 x 10^-26) in 309 DD/ID genes. Six genes are shared with BipEx 2.0 bipolar results. Six of 40 genes lie within 500 kb of 287 GWAS index SNPs (empirical p = 0.41, 100,000 permutations); in 64 fine-mapped GWAS genes, damaging missense OR = 1.28 (p = 6.45e-6). · Overlap Across Neuropsychiatric Genes; Shared GWAS signal

  5. ShownAll 40 genes are relatively more expressed in excitatory and inhibitory neurons, and brain regions dominate the tissue enrichment. Their developmental expression peaks range from prenatal to adult, and synapse-related gene ontology terms are enriched.

    11 of the top 12 GTEx tissues (via FUMA) are brain regions. PsychENCODE prefrontal snRNA-seq (294 donors): all genes elevated in neurons on within-gene z-scores; k-means gives clusters of 23 neuronal, seven oligodendrocyte-lineage, five astrocyte and four microglial genes. BrainSpan: prenatal-peak (e.g. ZMYM2, SP4) and adolescent/adult-peak (e.g. SCN2A) clusters. g:Profiler: synapse is the top cellular component. · Gene Expression Patterns; Cellular Functions

Methods, data and reproducibility

Per-gene Cochran-Mantel-Haenszel burden tests, stratified by ancestry and exome capture kit, compared ultra-rare (MAC ≤ 15) PTVs alone, and PTVs plus MisMeanRank ≥ 93 missense variants, between QC-matched cases and controls from 120 cohorts. The genes were then checked for cross-disorder overlap, GWAS proximity, brain expression, gene ontology enrichment and phenome-wide associations in All of Us carriers.

  • SCHEMA 2.0 exome case-control set (120 cohorts, five capture methods; includes SCHEMA 1.0, gnomAD v4 schizophrenia samples and new blended genome exome data) · human · n = 87,959 cases and 150,587 controls after QC
  • De novo mutations from SCHEMA 1.0 schizophrenia trios · human · n = 3,402 trios
  • GTEx bulk RNA-seq tissue expression, via FUMA GENE2FUNC · human (GTEx)
  • PsychENCODE aggregated dorsolateral prefrontal cortex single-nucleus RNA-seq, via PsychScreen · human (donors with and without psychiatric disorders) · n = 294 donors across eight cohorts
  • BrainSpan developmental brain transcriptome, seven stages from first trimester to adulthood · human (BrainSpan; prenatal trimesters to adulthood)
  • All of Us Research Program Controlled Tier Dataset v8 (whole-genome sequencing plus electronic health records) · human · n = 307,694 individuals; 1,581 PTV carriers
  • Latest schizophrenia GWAS: independent loci (GRCh38) and fine-mapped protein-coding genes · human · n = 287 loci; 64 fine-mapped genes
  • Published gene sets for DD/ID (Kaplanis et al. 2020), ASD (Fu & Satterstrom et al.) and bipolar disorder (BipEx 2.0) · human · n = 309 DD/ID genes; 253 ASD genes
Methods & tools
Ensembl VEP v95, LOFTEE, MisMeanRank, MPC, AlphaMissense, MisFit-S, PopEVE, gnomAD v4, blended genome exome (BGE) sequencing, FUMA GENE2FUNC, PsychScreen, g:Profiler
Software & environment
Variant annotation with Ensembl VEP v95 and LOFTEE; gnomAD v4 frequencies and gnomAD v2 mutation rates; All of Us Controlled Tier Dataset v8. No software versions, containers or requirement files are given for the analysis code.
Reproducibility gaps
Release of individual-level sequence data is not described; the browser shares variant-level and gene-level results. The All of Us data are Controlled Tier. Detailed QC is in Supplementary Methods, which are not included in this text.
Limitations (6)
  • Case ascertainment and control matching differ across cohorts, and many diagnoses were recorded decades ago, so diagnostic stability and illness course cannot be assessed. · stated by the authors
  • Sequencing technology changed over the project (five capture methods, five callsets). The authors address this with harmonized QC and association pipelines. · stated by the authors
  • Few carriers per gene leave odds ratio intervals wide and results unstable. GRIN2A, GRIA3 and CACNA1G fell below 5% FDR, possibly because of stricter QC, new cohorts less enriched for comorbid DD/ID, or false positives in SCHEMA 1.0. · stated by the authors
  • The genes are not diagnostic: controls also carry damaging variants, and fewer than 2% of cases carry a damaging mutation in an identified gene. The analysis also ignores compensatory expression from the reference allele. · stated by the authors
  • MisMeanRank's component scores were chosen for schizophrenia enrichment (GeneticsGym), and its 93rd-percentile cutoff was tuned on case-control enrichment within the non-gnomAD part of this dataset. Its advantage over MPC on the same data may therefore be optimistic. · noted in this summary
  • No independent replication cohort is reported for the 40 genes. SCHEMA 1.0 samples are part of SCHEMA 2.0, so the persistence of SCHEMA 1.0 genes is not independent confirmation. · noted in this summary

Read this if you want the current schizophrenia rare-variant gene list with effect sizes and carrier counts, or are choosing missense annotations and stratification for an exome burden study.

Citations
1

Abstract

In this study, we present the largest rare coding variant association study of schizophrenia to date, with 87,959 schizophrenia cases and 150,587 controls. We identify 16 genes at exome-wide significance: SETD1A, ZMYM2, HERC1, RB1CC1, SCAF1, XPO7, SP4, FYN, PPP3CA, CUL1, HDAC9, JARID2, ATP9A, PTK2, STAG1, and SCN2A, and an additional 24 at a 5% false discovery rate. Cases carrying ultra-rare, damaging variants, primarily protein-truncating and deleterious missense mutations, show strong enrichment in constrained genes with consistent effects across ancestries. Half of the 40 identified genes overlap with those implicated in developmental delay, autism, or bipolar disorder, underscoring shared neurodevelopmental risk. All 40 genes identified are highly expressed in excitatory and inhibitory neurons, and their expression patterns span diverse developmental stages. Functionally, the identified genes implicate chromatin regulation, protein degradation, and synaptic function. Overall, our results advance understanding of the genetic architecture of schizophrenia, provide a foundation for modeling risk genes in cellular and animal systems, and offer a framework for future work toward biologically informed patient stratification.

Reproducibility ledger

Nobody has reported trying to reproduce this paper yet. Did you replicate an experiment or run the code? Your report saves the next lab a week.

Sign in to file a reproducibility report.

Related papers

Similarity Map
SelectedSame fieldCross-field
THISSingh '16Singh '22Chick '24Trubetskoy '22Lencz '21Akingbuwa '22
84% semantic matchGenetics
Tarjinder Singh, James T. R. Walters, Mandy Johnstone+18 more·bioRxiv·Aug 16, 2016