Biology subjects

Chen, Z.-Y.

Publications and source records attributed to Chen, Z.-Y..

11 recordsLinked to original sources

bioRxiv2026

Somatic haplotype reconstruction and variant recalibration from tumor-only long-read sequencing

Separating somatic from germline variants and reconstructing somatic haplotypes are the two central problems of tumor-only cancer genome analysis. Long reads carry the linkage needed to solve both, but chromosome-scale loss of heterozygosity (LOH) and an unknown degree of normal-cell admixture blur the distinction between somatic and germline haplotypes. Here we present LongPhase-TO, the first method to reconstruct somatic haplotypes from a tumor sample alone. Rather than mapping somatic variants onto germline haplotypes, LongPhase-TO co-phases germline and somatic alleles in a unified graph, in which LOH and tumor DNA fraction are resolved internally from heterozygosity depletion and haplotype imbalance rather than a copy-number and ploidy model. Across eight datasets from six cancer cell lines, LongPhase-TO increased haplotype block N50 by a median of 2.9-fold relative to germline phasers. It also consistently improved somatic single-nucleotide variant (SNV) and indel calls from ClairS-TO and DeepSomatic-TO, raising mean F1 from 0.55 to 0.62 and 0.65 for SNVs and from 0.19 to 0.23 for indels, with the largest gains at low tumor DNA fraction. Across breast, melanoma and lung cancer cell lines, LongPhase-TO improves the accuracy of existing somatic callers and reconstructs megabase-scale somatic haplotypes.

bioinformatics↗

bioRxiv2026

Relational biological structure improves fine-mapping of causal GWAS variants under weak signal

Linkage disequilibrium (LD) makes causal GWAS variants indistinguishable from correlated neighbours; resolving them is the fine-mapping problem, and the challenge is species-specific: humans face dense ancestry-imbalanced LD, yeast and Arabidopsis exceptionally long LD, and crop germplasm sparse and fragmented annotations that defeat human-biobank curation pipelines. Bayesian fine-mappers integrate annotations as flat per-variant priors, discarding the relational structure linking variants to tissue-specific eQTLs, pathways and protein-protein interactions. Hierarchical belief propagation (HBP) on a variant-gene-pathway factor graph matches Bayesian baselines at 5-40x speed; an annotation-adaptive complement, graph-augmented fine-mapping (GAFM), wins 27-2 against SuSiE at weak signal and recovers LDLR, APOE, LPL, GCKR and ANGPTL3 at single-variant resolution across four Pan-UK Biobank ancestries. On the 3,000 Rice Genomes grain weight + shape panel, mixture-prior posterior reweightings of GAFM/HBP and their ensemble (GAFM-MX, HBP-MX, ENS) reach 47.6% top-1-PIP exact-position recovery of 21 panel-matched stable QTNs -- the highest of any method, exceeding SuSiE (28.6%) and SBayesRC (14.3%) -- at 200-700x SuSiEs per-locus speed. Across 692 leads in four species, a non-uniform per-variant prior, not uniform high coverage, lets the graph break LD ties: adding a regulatory-element flag to an otherwise uniform human cache flips HBP narrower than GAFM from 0% to 88% on 321 Pan-UKB leads. These results recast multi-omics fine-mapping as a non-uniform-prior-curation problem rather than a uniform-coverage problem, and reframe post-GWAS analysis as message passing over biological structure rather than weighted regression on flattened annotations.

genomics↗

bioRxiv2026

Canonical self-supervised pretraining paradigm constrains the capacity of genomic language models on regulatory decoding

Recent studies suggest that genomic language models (gLMs) could help decode genomic regulatory code. Here, we systematically evaluated 11 representative gLMs across multiple regulatory genomics applications and found that current gLMs offer limited advantages over the random baseline. Further analysis revealed a systematic misalignment between the canonical sequence-only self-supervised pretraining paradigm and the context-specific dynamic nature of gene regulation, highlighting the need for function-oriented pretraining strategies that explicitly incorporate biochemical and regulatory priors.

bioinformatics↗

bioRxiv2026

GraphMana: graph-native data management for population genomics projects

Population genomics projects rely on fragmented file-based workflows that lose provenance and require full reprocessing when samples are added. Graph-Mana stores variant data in a graph database as packed genotype arrays with pre-computed population statistics, enabling incremental sample addition, provenance tracking, cohort management, and export to 17 formats. Two access paths serve different needs: a FAST PATH reading population-level arrays in O(K) time and a FULL PATH unpacking per-sample genotypes in O(N) time. On human 1000 Genomes data (3,202 samples, 70.7M variants), Graph-Mana completed a 46-operation lifecycle in 98 minutes from a single persistent database.

bioinformatics↗

bioRxiv2026

GraphPop: graph-native computation decouples population genomics complexity from sample count

Matrix-based population genomics tools scale as O(V x N ), re-reading the full genotype matrix for every analysis. Here we present GraphPop, a graph database engine that reduces summary statistic complexity to O(V x K) where K is population count--independent of sample count--by computing on pre-aggregated allele-count arrays stored as graph node properties. The same architecture enables annotation-conditioned queries via edge traversal, persistent analytical records, and multi-statistic composition. Applied to rice 3K (29.6M SNPs, 3,024 accessions) and human 1000 Genomes (3,202 samples, 22 autosomes), GraphPop reveals that all 12 rice subpopulations show{pi} N /{pi}S > 1.0, uncovers opposite consequence-level Fst regimes between species, and identifies KCNE1 as a candidate pre-Out-of-Africa sweep via convergence of five stored statistics. GraphPop achieves 146-327x query-time speedup for pre-aggregated statistics and 63-179x for bit-packed haplotype computation (iHS, XP-EHH, nSL), at constant[~] 160 MB procedure working memory. This complexity reduction makes systematic, annotation-integrated population genomics practical at scale.

bioinformatics↗

bioRxiv2024

CD388: A universally protective Drug-Fc Conjugate that targets influenza virus neuraminidase

The ability of the influenza virus to elude humoral immunity by rapid antigenic shift presents a sustained and urgent threat to human health. Globally, seasonal influenza causes an estimated 3 - 5 million cases of severe disease and 300,000 - 500,000 deaths annually, with increased potential for mortality during pandemics1. The recent outbreaks of avian H5N1 in bird populations, subsequent spread to cattle and appearance in humans, highlight the need for effective broad-spectrum influenza antivirals for treatment, prophylaxis and pandemic preparedness. The narrow, strain-specific immunity induced by current seasonal vaccines and mismatches between vaccine strains and circulating viruses result in limited vaccine effectiveness (VE) rates2. Furthermore, VE is even lower in immune-compromised and - senescent populations3. Strategies that provide durable, universal influenza protection in healthy and high-risk populations are urgently needed. Here, we describe CD388, a first-in-class antiviral drug-Fc conjugate (DFC) in clinical trials for seasonal influenza prevention (NCT05285137 and NCT05523089). CD388 comprises a multivalent small molecule inhibitor of influenza virus neuraminidase (NA) linked to a CH1-Fc hybrid domain of human IgG1 engineered for extended half-life. CD388 demonstrated potent, universal activity across influenza A and B viruses, including high pathogenicity and NA resistant strains, a low potential for resistance development, and efficacy in lethal mouse infection models. CD388 is the first therapeutic with the potential for universal prevention of influenza A and B in healthy and high-risk populations.

microbiology↗

bioRxiv2023

Targeted genome editing restores auditory function in adult mice with progressive hearing loss caused by a human microRNA mutation

Mutations in microRNA-96 (MIR96) cause dominant delayed onset hearing loss DFNA50 without treatment. Genome editing has shown efficacy in hearing recovery by intervention in neonatal mice, yet editing in the adult inner ear is necessary for clinical applications. Here, we developed an editing therapy for a C>A point mutation in the seed region of the Mir96 gene, Mir9614C>A associated with hearing loss by screening gRNAs for genome editors and optimizing Cas9 and sgRNA scaffold for efficient and specific mutation editing in vitro. By AAV delivery in pre-symptomatic (3-week-old) and symptomatic (6-week-old) adult Mir9614C>A mutant mice, hair cell on-target editing significantly improved hearing long-term, with an efficacy inversely correlated with injection age. We achieved transient Cas9 expression without the evidence of AAV genomic integration to significantly reduce the safety concerns associated with editing. We developed an AAV-sgmiR96-master system capable of targeting all known human MIR96 mutations. As mouse and human MIR96 sequences share 100% homology, our approach and sgRNA selection for efficient and specific hair cell editing for long-term hearing recovery lays the foundation for future treatment of DFNA50 caused by MIR96 mutations.

bioengineering↗

bioRxiv2023

Preclinical evaluation of the efficacy and safety of AAV1-hOTOF in mice and non-human primates

Pathogenic mutations in the OTOF gene cause autosomal recessive hearing loss 9 (DFNB9), one of the most common forms of auditory neuropathy. There is no biological treatment for DFNB9. Here, we designed an OTOF gene therapy agent by dual AAV1 carrying human OTOF coding sequences with the expression driven by the hair cell-specific promoter Myo15, AAV1-hOTOF. To develop a clinical application of AAV1-hOTOF gene therapy, we evaluated its efficacy and safety in animal models by pharmacodynamics, behavior, and histopathology. AAV1-hOTOF inner ear delivery significantly improved hearing in Otof -/- mice without affecting normal hearing in wild-type mice. AAV1 was predominately distributed to the cochlea although it was detected in other organs such as the central nervous system and the liver, and no obvious toxic effects of AAV1-hOTOF were observed in mice. To further evaluate the safety of Myo15 promoter-driven AAV1-transgene, AAV1-GFP was delivered into the inner ear of Macaca fascicularis via the round window membrane. AAV1-GFP transduced 60-94% of the inner hair cells along the cochlear turns. AAV1-GFP was detected in isolated organs and no significant adverse effects were detected. These results suggest that AAV1-hOTOF is well tolerated and effective in animals, providing critical support for its clinical translation.

genetics↗

bioRxiv2023

Rescue of Auditory Function by a Single Administration of AAV-TMPRSS3 Gene Therapy in Aged Mice of Human Recessive Deafness DFNB8

AbstractsPatients with mutations in the TMPRSS3 gene suffer from recessive deafness DFNB8/DFNB10 for whom cochlear implantation is the only treatment option. Poor cochlear implantation outcomes are seen in some patients. To develop biological treatment for TMPRSS3 patients, we generated a knock-in mouse model with a frequent human DFNB8 TMPRSS3 mutation. The Tmprss3A306T/A306T homozygous mice display delayed onset progressive hearing loss similar to human DFNB8 patients. Using AAV2 as a vector to carry a human TMPRSS3 gene, AAV2-hTMPRSS3 injection in the adult knock-in mouse inner ears results in TMPRSS3 expression in the hair cells and the spiral ganglion neurons. A single AAV2-hTMPRSS3 injection in aged Tmprss3A306T/A306T mice leads to sustained rescue of the auditory function, to a level similar to the wildtype mice. AAV2-hTMPRSS3 delivery rescues the hair cells and the spiral ganglions. This is the first study to demonstrate successful gene therapy in an aged mouse model of human genetic deafness. This study lays the foundation to develop AAV2-hTMPRSS3 gene therapy to treat DFNB8 patients, as a standalone therapy or in combination with cochlear implantation.

neuroscience↗

bioRxiv2020

Degradation of SARS-CoV-2 receptor ACE2 by tobacco carcinogen-induced Skp2 in lung epithelial cells

An unexpected observation among the COVID-19 pandemic is that smokers constituted only 1.4-18.5% of hospitalized adults, calling for an urgent investigation to determine the role of smoking in SARS-CoV-2 infection. Here, we show that cigarette smoke extract (CSE) and carcinogen benzo(a)pyrene (BaP) increase ACE2 mRNA but trigger ACE2 protein catabolism. BaP induces an aryl hydrocarbon receptor (AhR)-dependent upregulation of the ubiquitin E3 ligase Skp2 for ACE2 ubiquitination. ACE2 in lung tissues of non-smokers is higher than in smokers, consistent with the findings that tobacco carcinogens downregulate ACE2 in mice. Tobacco carcinogens inhibit SARS-CoV-2 Spike protein pseudovirions infection of the cells. Given that tobacco smoke accounts for 8 million deaths including 2.1 million cancer deaths annually and Skp2 is an oncoprotein, tobacco use should not be recommended and cessation plan should be prepared for smokers in COVID-19 pandemic.

cell biology↗

bioRxiv2020

Exon 13-skipped USH2A protein retains functional integrity in mice, suggesting an exon skipping therapeutic approach to treat USH2A-associated disease.

Mutations in the USH2A gene are the most common cause of non-syndromic inherited retinal degeneration and Usher syndrome, which is characterized by congenital deafness and progressive vision loss. Development of a vector mediated therapy for USH2A-associated disease has been challenging due to its large size of coding sequence (~15.6kb). Therefore, there is an unmet need to develop alternative therapeutic strategies. The USH2A protein (Usherin) contains many repetitive domains, and it has been hypothesized that some domains may be dispensable with regard to protein function. Here, we show that skipping of exon 13 of the human USH2A gene or the equivalent exon 12 of the mouse Ush2a gene results in an in-frame transcript that produces functional Usherin protein. This nearly full length Usherin rescues the ciliogenesis in Ush2a null cells as well as the cochlear and retinal phenotypes in Ush2a null mice. Together, our results support the development of exon-skipping strategies to treat both visual and hearing loss in patients with USH2A-associated disease due to mutations in exon 13.

molecular biology↗