Konrad J. Karczewski

The paper reports on the All of Us multi-omics cohort, identifying 182,246 QTLs and demonstrating rare variant effects across proteomic and transcriptomic data.

PESTO classifies gene-phenotype associations as Novel, Hypothesized, Existing, or Established using a large language model and structured genetic evidence, identifying 7% as Novel and 21% as Hypothesized in 940 associations.

ALLSPICE identifies heterogeneous rare variant effects across traits in gene-trait associations using summary statistics from the UK Biobank.

The paper presents a large reference panel of 515,579 genomes to improve rare-variant imputation across underrepresented populations, enhancing accuracy for variants with frequencies as low as 0.002%.

This study identifies BAF complex interaction with FOS in regulating ASD-associated gene programs through enhancer activation and shows how genetic variants disrupt this process.

The paper estimates the contribution of de novo coding mutations to autism using data from 38,680 trios, finding they explain 3.4% of autism variance.

The paper identifies 253 autism-associated genes using rare variation from 62,429 individuals and finds pleiotropy across physiological systems and developmental disorders.

FELIX improves genetic association analysis by accounting for local ancestry in admixed populations, increasing significant loci and improving polygenic score prediction.

The paper identifies abundant alternate translation products in human proteins, showing their association with protein stability and disease.

The paper identifies 260 genes associated with sleep phenotypes using rare coding variants from large-scale sequencing data, revealing distinct genetic architectures and biological mechanisms across different sleep measurements.

The paper estimates mutation rates and variant effects using 700k exomes to identify genes with mutational hotspots and improve variant prioritization for genetic diagnostics.

The paper develops a framework to map rare variants onto protein structures and identifies disease-associated 3D neighborhoods in 872 genes for Mendelian disease and others for autism, epilepsy, and schizophrenia.

The paper establishes a global consortium to discover rare genetic associations by integrating data from multiple biobanks, identifying 514 gene-trait associations, many detectable only through cross-ancestry meta-analysis.

The paper performs large-scale GWAS and RVAS on 392,030 whole genomes from All of Us, identifying 49,863 associations and 193 novel pLoF gene-phenotype links through meta-analysis with UK Biobank.

Cell Painting in primary human hepatocytes exposed to 1,085 compounds detects bioactivity at lower concentrations than cytotoxicity assays and predicts cell-based, but not cell-free, ToxCast endpoints across feature extractors.

gnomAD v4 (730,947 exomes) brings LOFTEE-2 loss-of-function annotation (0.90 precision, strict mode), missense-aware constraint, and LLM-mined literature evidence combined with constraint to rank disease genes.

The paper identifies naturally variable exons in human transcripts, showing their prevalence and impact on gene expression and genetic interpretation.

This study identifies 12 enhancers in the 9p21.3 CAD risk locus that regulate MTAP, modulating vascular cell-state transitions and CAD risk through stress-responsive mechanisms.

This paper improves allele frequency estimates in gnomAD by applying local ancestry inference to admixed populations, revealing significant frequency differences that could affect clinical interpretation.

The paper investigates the mechanism of age-related mtDNA mutation accumulation in human blood using whole genome sequences, revealing a two-step process involving cryptic mutations and clonal expansion.

The paper introduces VCF Zarr, a format for efficiently storing and processing large-scale genetic variation data, demonstrating improved performance and scalability over traditional VCF formats.

This paper introduces SVCR, a scalable variant call format that reduces file size and enables efficient analysis of large genetic datasets.

This study evaluates variant scoring methods across different selection regimes, showing their performance depends on variant-to-gene and gene-to-disease factors.

The paper harmonizes diverse human genomes from multiple datasets to create a new reference panel with improved variant coverage and population structure analysis.

The paper quantifies regional missense mutational intolerance using 730,947 exomes to identify constrained gene regions and improve variant interpretation.

Meta-analyzing per-ancestry GWAS of 7,266 UK Biobank traits across six genetic ancestry groups adds 14,676 loci missed by European-only analysis, including ancestry-enriched CAMK2D and G6PD signals.

CHARR estimates DNA contamination from variant-level sequencing data, offering accurate results with reduced computational costs compared to existing methods.

This paper presents a framework to reduce false positives in predicted loss-of-function variants by analyzing LoF evasion mechanisms and adjusting ACMG/AMP guidelines.

Blood whole genomes from 274,832 people yield 92 independent nuclear signals for mtDNA copy number and 42 for inherited heteroplasmy levels, implicating mtDNA replication machinery.

The paper identifies a genotyping artifact causing apparent HWE deviation at CCR5-Δ32 and other variants, showing it results from contamination rather than selective pressure.