Chunyu Liu

The paper identifies distinct DNA methylation mechanisms for common and rare noncoding genetic risk in brain disorders using deep learning and single-nucleus DNAm data.

The paper measures 5-methyl-cytosine and 5-hydroxymethyl-cytosine in human neurons across the lifespan, showing dynamic changes in hmC linked to transcription, chromatin state, and sex-specific gene regulation.

The paper identifies eQTLs, rQTLs, and pQTLs in the human brain and shows that many eQTL effects are attenuated post-transcriptionally, yet still correlate with schizophrenia GWAS signals.

This study evaluates precision and accuracy in sc/snRNA-seq data, identifies factors affecting reproducibility, and introduces a tool to assess data quality and differential expression reliability.

The paper analyzes psychosis biotypes using ancestry-adjusted polygenic risk scores and identifies genomic associations with specific biotypes in brain transcriptomes.

This study identifies population-specific eQTLs in brain tissues from European, African American, and Asian ancestry individuals, revealing new schizophrenia risk genes and variants missed in European-only datasets.

The paper identifies extensive isoform diversity in the developing human neocortex and links it to neuropsychiatric disorder risk mechanisms.

The paper evaluates deconvolution methods on human brain data, finding dtangle and bMIND perform best, and identifies decon-eQTLs that explain more schizophrenia heritability than bulk or single-cell eQTLs.

The paper introduces a method for deconvolving bulk proteomes using transcriptome-proteome integration and applies it to identify cell-type-specific protein quantitative trait loci.

The paper re-analyzes RNA sequencing data comparing gene expression in living and postmortem human brain tissue, finding that living samples had lower quality, and questions the validity of previous conclusions about postmortem studies.

The paper constructs a cross-ancestry, cell-type-informed gene regulatory atlas of the developing human brain, identifying genetic regulation and linking it to neuropsychiatric disorders.

This study conducts GWAS on 11 mathematical ability categories in Chinese children, identifying 7 genome-wide significant SNPs and replicating one SNP associated with division ability.

The paper investigates the impact of the NLGN3 R451C mutation on human neurons, finding it reduces protein levels but enhances excitatory synapses, potentially contributing to ASD pathogenesis.

The paper identifies MAPK3 as a driver gene in 16p11.2 deletion-associated neurodevelopmental disorders, showing haplotype-specific expression and cis-regulatory enhancer SNPs contribute to variable phenotypes.

swCAM estimates subtype-specific expressions in individual samples using unsupervised deconvolution, revealing changes in cell proportions and gene expression in bipolar disorder.

The paper identifies correlated DNA methylation and gene expression pairs across tissues and brain development stages, finding most are tissue-specific and revealing novel regulatory mechanisms.

SMAP is a pipeline that matches proteomic and genomic samples using protein-coding variants and discriminant scores, correcting 18.8% of mismatches in a large-scale proteomics dataset.

This paper introduces an unsupervised deconvolution method, CAM3.0, to determine molecular archetype composition and expression from bulk tissues, showing its effectiveness through simulations and case studies.

The paper compiles and analyzes transcriptomic data from six neuropsychiatric disorders, identifying differentially expressed genes and co-expression networks across brain regions.

This paper identifies potential nootropic drug targets by integrating large-scale cognitive GWAS and transcriptomic data to find credible druggable genes.

This study identifies fever and infection as major agonal factors affecting gene expression in postmortem brains and recommends correcting for these factors using PEER or SVA.

This study identifies eQTL and sQTL in East Asian brains, finds convergence in transcriptional regulatory architecture with European populations, and links these to schizophrenia risk genes and pathways.

The paper uses cell-type deconvolution to analyze gene expression in schizophrenia and finds upper-layer neurons in the DLPFC are associated with SCZ risk.

DRAMS detects and corrects sample mix-ups in multi-omics data by aligning genetic and omics data, improving statistical power and result quality.

csuWGCNA combines signed and unsigned WGCNA to better detect negative correlations, improving module detection and biological interpretation in gene regulation.

The paper evaluates brain cell marker genes using differential and co-expression data, identifying 22 gold standard and 16 novel candidates with improved specificity.

The paper investigates the glia-neuron ratio in the brain using unsupervised deconvolution methods on single-cell data from human brains.

The paper identifies positional effects in Illumina Methylation Arrays and shows how they impact methylation measurements, recommending ComBat for correction.

The paper reports GWAS findings linking common variants in NRXN1, LRP1B, and RORA to increased ventricular volumes in psychosis, using deep phenotyping data from 1,115 cases and controls.