A DNA language model based on multispecies alignment predicts the effects of genome-wide variants

Nature Biotechnology, Journal Year: 2025, Volume and Issue: unknown

Published: Jan. 2, 2025

Language: Английский

---

Macroevolutionary divergence of gene expression driven by selection on protein abundance

Science, Journal Year: 2025, Volume and Issue: 387(6738), P. 1063 - 1068

Published: March 6, 2025

The regulation of messenger RNA (mRNA) and protein abundances is well-studied, but less known about the evolutionary processes shaping their relationship. To address this, we derived a new phylogenetic model applied it to multispecies mammalian data. Our analyses reveal (i) strong stabilizing selection on over macroevolutionary time, (ii) mutations affecting mRNA minimally impact abundances, (iii) evolve under align with (iv) adapt faster than owing greater mutational opportunity. These conclusions are supported by comparisons parameters independent functional genomic By decomposing selective influences mRNA-protein dynamics, our approach provides framework for discovering rules that drive divergence in gene expression.

Language: Английский

---

Proteome-scale prediction of molecular mechanisms underlying dominant genetic diseases

PLoS ONE, Journal Year: 2024, Volume and Issue: 19(8), P. e0307312 - e0307312

Published: Aug. 22, 2024

Many dominant genetic disorders result from protein-altering mutations, acting primarily through dominant-negative (DN), gain-of-function (GOF), and loss-of-function (LOF) mechanisms. Deciphering the mechanisms by which diseases exert their effects is often experimentally challenging resource intensive, but essential for developing appropriate therapeutic approaches. Diseases that arise via a LOF mechanism are more amenable to be treated conventional gene therapy, whereas DN GOF may require editing or targeting small molecules. Moreover, pathogenic missense mutations act difficult identify than those using nearly all currently available variant effect predictors. Here, we introduce tripartite statistical model made up of support vector machine binary classifiers trained predict whether human protein coding genes likely associated with DN, GOF, molecular disease We test utility predictions examining biologically clinically meaningful properties known Our results strongly models able generalise on unseen data offer insight into functional attributes proteins different hope our will serve as springboard researchers studying novel variants uncertain clinical significance, guiding interpretation strategies experimental characterisation. Predictions UniProt reference proteome at https://osf.io/z4dcp/ .

Language: Английский

---

Improved multi-ancestry fine-mapping identifiescis-regulatory variants underlying molecular traits and disease risk

medRxiv (Cold Spring Harbor Laboratory), Journal Year: 2024, Volume and Issue: unknown

Published: April 16, 2024

Abstract Multi-ancestry statistical fine-mapping of cis -molecular quantitative trait loci ( -molQTL) aims to improve the precision distinguishing causal -molQTLs from tagging variants. However, existing approaches fail reflect shared genetic architectures. To solve this limitation, we present Sum Shared Single Effects (SuShiE) model, which leverages LD heterogeneity precision, infer cross-ancestry effect size correlations, and estimate ancestry-specific expression prediction weights. We apply SuShiE mRNA measured in PBMCs (n=956) LCLs (n=814) together with plasma protein levels (n=854) individuals diverse ancestries TOPMed MESA GENOA studies. find fine-maps for 16 % more genes compared baselines while prioritizing fewer variants greater functional enrichment. infers highly consistent -molQTL architectures across on average; however, also evidence at predicted loss-of-function intolerance, suggesting that environmental interactions may partially explain differences sizes ancestries. Lastly, leverage estimated effect-sizes perform individual-level TWAS PWAS six white blood cell-related traits AOU Biobank (n=86k), identify 44 baselines, further highlighting its benefits identifying relevant complex disease risk. Overall, provides new insights into -genetic architecture molecular traits.

Language: Английский

---

Gene regulatory network inference from CRISPR perturbations in primary CD4+ T cells elucidates the genomic basis of immune disease

Cell Genomics, Journal Year: 2024, Volume and Issue: 4(11), P. 100671 - 100671

Published: Oct. 11, 2024

Language: Английский

---

Bayesian predictive system for assessing the damage intensity of residential masonry buildings under the impact of continuous ground deformation

Scientific Reports, Journal Year: 2025, Volume and Issue: 15(1)

Published: Jan. 9, 2025

Abstract The paper introduces a method for predicting damage intensity in masonry residential buildings situated mining areas, focusing on the impact of large-scale continuous ground deformation. research utilizes situ data collected database, encompassing structural and material features, as well information maintenance quality building durability. In addition to this information, database deformation area at location building, range identified buildings. included was result many years observations during disclosure impacts from exploitation based on: results in-situ inventory, analysis available documentation provided by companies. archived were categorized variables labeled. transformation labeled value dictated directly assumptions GOBNILP algorithm. Ultimately, predictive model, represented an optimal Bayesian network structure, is established. optimisation structure achieved through adaptation learning algorithm data. This process executed Gurobi Optimizer. It worth noting that interdisciplinary approach represents one first applications such methodology field civil environmental engineering. obtained can therefore be significant given fact detecting networks still developing intensively other scientific fields. course analyses, metric scores are examined, various structures assessed their complexity. Great values classification accuracies over 91% obtained. meticulous evaluation allows selection best generalises knowledge acquired process. also demonstrates potential application model diagnosing causes future occurrences, highlighting versatility proposed addressing issues field.

Language: Английский

---

Causal modeling of gene effects from regulators to programs to traits: integration of genetic associations and Perturb-seq

bioRxiv (Cold Spring Harbor Laboratory), Journal Year: 2025, Volume and Issue: unknown

Published: Jan. 24, 2025

Genetic association studies provide a unique tool for identifying causal links from genes to human traits and diseases. However, it is challenging determine the biological mechanisms underlying most associations, we lack genome-scale approaches inferring mechanistic pathways cellular functions traits. Here propose new bridge this gap by combining quantitative estimates of gene-trait relationships loss-of-function burden tests with gene-regulatory connections inferred Perturb-seq experiments in relevant cell types. By these two forms data, aim build graphs which directional associations trait can be explained their regulatory effects on programs or direct trait. As proof-of-concept, constructed graph gene hierarchy that jointly controls three partially co-regulated blood We perturbation trait-relevant types, coupled gene-level effect sizes traits, between genetics biology.

Language: Английский

---

Mapping the regulatory effects of common and rare non-coding variants across cellular and developmental contexts in the brain and heart

bioRxiv (Cold Spring Harbor Laboratory), Journal Year: 2025, Volume and Issue: unknown

Published: Feb. 19, 2025

Abstract Whole genome sequencing has identified over a billion non-coding variants in humans, while GWAS revealed the as significant contributor to disease. However, prioritizing causal common and rare human disease, understanding how selective pressures have shaped genome, remains challenge. Here, we predicted effects of 15 million with deep learning models trained on single-cell ATAC-seq across 132 cellular contexts adult fetal brain heart, producing nearly two context-specific predictions. Using these predictions, distinguish candidate underlying traits diseases their effects. While variant are more cell-type-specific, exert cell-type-shared regulatory effects, particularly targeting affecting neurons. To prioritize de novo mutations extreme developed FLARE, functional genomic model constraint. FLARE outperformed other methods case from autism-affected families near syndromic autism-associated genes; for example, identifying mutation outliers CNTNAP2 that would be missed by alternative approaches. Overall, our findings demonstrate potential integrating maps population genetics learning-based effect prediction elucidate mechanisms development disease–ultimately, supporting notion genetic contributions neurodevelopmental disorders predominantly rare.

Language: Английский

---

Gene dosage architecture across complex traits

medRxiv (Cold Spring Harbor Laboratory), Journal Year: 2025, Volume and Issue: unknown

Published: Feb. 26, 2025

Copy number variants (CNVs) have large effects on complex traits, but they are rare and remain challenging to study. As a result, our understanding of biological functions linking gene dosage traits remains limited, whether these sensitive similar those underlying the single nucleotide (SNVs) common unknown. We developed FunBurd, functional burden analysis, test association CNVs aggregated within sets. applied this approach in 500,000 individuals from UK Biobank associate 43 with disrupting 172 sets across tissues cell types. compared CNV findings LoF (Loss Function) SNVs same cohort using All showed FDR significant associations CNVs. Brain tissue neuronal cell-types highest levels pleiotropy. Most set could, part, be explained by genetic constraint, except for brain related processes. Shared contributions between pairs were concordant types variants, average 2-fold higher, variants.Functional enrichment found limited overlap variants. Moreover, deletions duplications negatively correlated most traits.In conclusion, we present new methods separate constraint function traits. Overall, convergence different -even duplications- limited. limited.FunBurd (functional analysis) was UKBiobank tissues/cell type sets.All brain-related higher The Our provide insights

Language: Английский

---

Functionally constrained human proteins are less prone to mutational instability from single amino acid substitutions

Nature Communications, Journal Year: 2025, Volume and Issue: 16(1)

Published: March 13, 2025

Abstract Missense mutations that disrupt protein structural stability are a common pathogenic mechanism in human genetic disease. Here, we quantify potential disruption of due to amino acid substitution and show functionally constrained proteins less susceptible large mutational changes stability. Mechanistically, this relates greater intrinsic disorder among increased B-factors the ordered regions proteins. This phenomenon means exhibit smaller effects missense mutations, partly explains why overtransmission variation is prevalent disorders characterised by truncations. We most depleted both destabilising overly-stabilising disease-free populations. Despite this, substitutions with still highly variation. Importantly, observe there approximately five times more variants than unambiguous loss-of-function mutations. recapitulate per-gene patterns functional constraint observed truncating variation, yet their relative abundance abrogates difficulties encountered when estimating for shortest genes.

Language: Английский

---