Browse Articles

An analysis of UK Biobank participants shows that the risk of developing different types of myeloid neoplasms can be inferred years before diagnosis. The authors integrate somatic gene mutations with blood test parameters into a predictive model, which could guide future strategies for early detection and prevention of these diseases.

GATK-gCNV uses a probabilistic model and inference framework to discover rare copy number variants (CNVs) from sequencing read-depth information. This algorithm is used to generate a reference catalog of rare coding CNVs in exome sequencing data from UK Biobank.

Meta-analysis of three large whole-exome sequencing datasets highlights protein-truncating and rare missense variants associated with breast cancer susceptibility.

Genome-wide association analyses of magnetic resonance imaging data describe the genetic architecture of 13 cortical phenotypes at both global and regional levels, implicating neurodevelopmental and constrained genes.

Bulk ex vivo and single-cell in vivo CRISPR knockout screens are used to characterize 680 chromatin factors during mouse hematopoiesis, highlighting lineage-specific and normal and leukemia-specific functions.

Tissue co-regulation score regression (TCSC) infers causal tissues and partitions trait heritability into tissue-specific components using a transcriptome-wide association study framework. Applying TCSC to 78 complex traits and diseases identifies biologically plausible tissue–trait relationships.

A novel pipeline that expands the utility of the protein language model ESM1b has provided variant effect predictions for more than 40,000 protein isoforms. This strategy outperformed several state-of-the-art methods over multiple benchmarks.

A modified framework leveraging a protein language model (ESM1b) is used to predict all possible 450 million missense variant effects in the human genome and shows potential for generalizing to more complex genetic variations such as indels and stop-gains.

Analysis of exome sequencing data identifies a burden of rare coding variants in 19 genes associated with bone mineral density. Integrated analyses show convergence of common- and rare-variant signals and highlight likely effector genes influencing osteoporosis risk.

Identifying genetic risk factors for binge-eating disorder (BED) is vital to understand its etiology and develop effective prevention and intervention strategies. To overcome under-reporting of clinical BED diagnosis, a new study uses machine learning to identify genetic variants associated with quantitative BED risk scores and finds evidence for a pathological role of heme metabolism.

Genome-wide association analysis of a binge eating disorder phenotype derived from a supervised machine-learning model applied to electronic medical records identifies three risk loci for this disorder and implicates iron metabolism in its etiology.

PhenoScore is an open-source machine-learning tool that combines facial image recognition with Human Phenotype Ontology for genetic syndrome identification without genomic data, with applications to subgroup analysis and variants of unknown significance classification.

Incidence of keratinocyte skin cancer varies markedly between populations living in different areas of the world. A detailed analysis of somatic mutations in the normal skin of individuals from the UK and Singapore reveals different patterns of clonal mutational landscapes that could contribute to differential risk.

A comparison of somatic mutations in skin from individuals from the UK and Singapore suggests that the difference in cancer incidence between the two countries is due to markedly different mutational spectra and patterns of selection.

Across multiple cancer types, hotspot mutations in SF3B1 confer selective sensitivity to multiple clinically available PARP inhibitors. This sensitivity is due to reduced levels of CINP specifically in SF3B1-mutant cells, which leads to a loss of the canonical replication stress response after challenge with PARP inhibitors.