Tianyuan Lu

CaPri: Bayesian Mendelian randomization for joint estimation of causal and pleiotropic effects

CaPri is a Bayesian Mendelian randomization framework that jointly estimates causal and pleiotropic effects, showing robust performance across varying pleiotropy proportions and real data applications.

Genomic atlas of cardiac and adiposity imaging phenotypes

The paper constructs a genomic atlas of cardiac and adiposity imaging phenotypes using whole-genome sequencing data from UK Biobank, identifying genetic associations and polygenic scores linked to cardiometabolic diseases.

Benchmarking Bayesian colocalization methods in validating Mendelian randomization-identified targets

This study benchmarks four Bayesian colocalization methods to validate MR-identified protein-trait associations for cardiometabolic traits, finding SharePro most frequently supports significant associations.

A multi-ancestry polygenic risk score for body mass index predicts longitudinal weight change

This paper develops a multi-ancestry polygenic risk score for BMI that predicts longitudinal weight change and shows strong associations across diverse genetic ancestries.

Single-cell gene programs define subtype identity and metastatic trajectories in renal cell carcinoma

This study identifies 59 gene expression programs in RCC that reveal subtype identity and metastatic trajectories, including a metastatic signature linked to poor clinical outcomes.

Genetic evidence prioritizes circulating proteins for heart failure beyond shared BMI-related genetic liability

This paper identifies nine circulating proteins associated with heart failure beyond the genetic component shared with BMI using GWAS-by-subtraction and proteome-wide Mendelian randomization.

Disentangling adiposity-related and non-adiposity-related genetic pathways for type 2 diabetes

This paper identifies 29 circulating proteins associated with type 2 diabetes risk independent of BMI using GWAS-by-subtraction, proteome-wide Mendelian randomization, and colocalization analyses.

Identifying circulating protein targets for common factors underlying schizophrenia, depression, and bipolar disorder

This study identifies 36 circulating proteins associated with a shared genetic liability for schizophrenia, depression, and bipolar disorder using genomic and proteomic data.

Colocalization and discordance between plasma and brain protein quantitative trait loci

The paper examines colocalization and discordance between plasma and brain cis-pQTLs, finding that most colocalize but some show opposite genetic effects, with implications for interpreting protein-trait associations.

Systematic variant-to-gene mapping highlights TGFB2 and VEGFA as adipokine-coding genes with non-obese, insulin-resistance-like characteristics and distinct disease risks

The paper maps adipokine-associated variants to effector genes across 1,669 traits, identifying TGFB2 and VEGFA as linked to insulin resistance and distinct disease risks.

Challenges to case-only analysis for gene-environment interaction detection using polygenic risk scores: model assumptions and biases in large biobanks

This study investigates challenges in case-only analysis using polygenic risk scores for gene-environment interaction detection, showing false positives due to model misspecification and collider bias in large biobanks.

Improving cross-ancestry generalizability of genetic risk prediction for short stature using a meta-polygenic risk score

This paper develops a meta-polygenic risk score (meta-PRS) for short stature by combining ancestry-specific PRS, improving cross-ancestry generalizability and predictive accuracy.

Multi-ancestry proteogenomic analysis identifies risk proteins for intracranial aneurysms

This paper identifies risk proteins for intracranial aneurysms using multi-ancestry proteogenomic analysis and finds ADH4 and PLEKHA1 as non-blood pressure-mediated candidates.

COSIME: Cooperative multi-view integration with Scalable and Interpretable Model Explainer

COSIME integrates multi-omics data to predict Alzheimer's disease phenotypes, improving accuracy and interpretability by analyzing feature interactions across data views.

Disentangling osteoarthritis-specific genetic effects from obesity to identify novel therapeutic targets

This study identifies OA-specific genetic effects and novel therapeutic targets by separating OA genetics from those shared with obesity using GWAS and proteomics data.

Integrative proteogenomic analyses provide novel interpretations of type 1 diabetes risk loci through circulating proteins

The paper uses proteogenomic data to identify circulating proteins associated with type 1 diabetes risk and explores their tissue-specific expression.

An accurate genetic colocalization method for the HLA locus

This paper introduces HLA-colocalization, a method for genetic colocalization at the HLA locus using HLA alleles, showing it identifies novel causal associations between viruses and autoimmune diseases.

MR Corge: Sensitivity analysis of Mendelian randomization based on the core gene hypothesis for polygenic exposures

MR Corge performs sensitivity analysis for Mendelian randomization by identifying core instruments to reduce horizontal pleiotropy bias and improve causal effect estimation.

Robust fine-mapping in the presence of linkage disequilibrium mismatch

RSparsePro improves fine-mapping accuracy in the presence of LD mismatch by using a probabilistic graphical model and variational inference, outperforming SuSiE in simulations and real applications.

Multi-ancestry proteome-phenome-wide Mendelian randomization offers a comprehensive protein-disease atlas and potential therapeutic targets

This study uses proteome-phenome-wide Mendelian randomization across three ancestries to identify protein-phenotype associations and potential therapeutic targets.

Valid inference for machine learning-assisted GWAS

This study identifies risks of false positives in ML-assisted GWAS and introduces POP-GWAS, a framework that provides valid inference for GWAS on ML-imputed phenotypes, applied to BMD data to find 89 novel loci.

Improved characterization of gene-environment interactions for vitamin D through variance quantitative trait loci

The paper identifies variance quantitative trait loci (vQTLs) for vitamin D levels and characterizes gene-environment interactions involving modifiable factors and sex.

A unified model for interpretable latent embedding of multi-sample, multi-condition single-cell data

This paper introduces GEDI, a generative model that unifies multi-sample, multi-condition single-cell analysis tasks, enabling cross-sample cell state mapping, differential gene expression, and pathway activity projection.

SharePro: an accurate and efficient genetic colocalization method accounting for multiple causal signals

SharePro improves genetic colocalization analysis by handling multiple causal signals, showing increased power and efficiency in simulations and real data.

Proteome-wide Mendelian randomization implicates nephronectin as an actionable mediator of the effect of obesity on COVID-19 severity

The paper identifies nephronectin as a mediator of obesity's effect on COVID-19 severity using proteome-wide Mendelian randomization and validates its expression in lung cells.

eXclusionarY: Ten years later, where are the sex chromosomes in GWAS?

The paper re-evaluates the inclusion of sex chromosomes in GWAS, finding persistent underrepresentation and proposing recommendations for improvement.

Transcriptional Reprogramming of Skeletal Muscle Stem Cells by the Niche Environment

This study shows that the niche environment significantly influences skeletal muscle stem cell transcriptional activity, with some age-related changes being reversible through exposure to a young niche.

HLA allele-calling using whole-exome sequencing identifies 129 novel associations in 11 autoimmune diseases: a multi-ancestry analysis in the UK Biobank

This study uses whole-exome sequencing to directly call HLA alleles in the UK Biobank, identifying 129 novel associations between HLA alleles and 11 autoimmune diseases.

A Sparse Additive Model for High-Dimensional Interactions with an Exposure Variable

The paper introduces a method called sail to detect non-linear interactions between a key exposure variable and high-dimensional data, showing it outperforms existing methods in prediction and support recovery.

Pan-cancer analysis of mRNA stability for decoding tumour post-transcriptional programs

The paper analyzes mRNA stability across 7760 cancer samples to identify stability changes and factors driving them in tumour development.