An analytical framework for decoding cell type-specific genetic variation of gene regulation

Question

1. How can single-cell RNA-sequencing (scRNA-seq) reveal cell type-specific eQTLs?

2. How does Huatuo framework analyze genetic regulation?

3. How do Huatuo predictions correlate with eQTL summary statistics?

4. What are cell cluster-ieQTLs and how do they provide insights into genetic regulatory effects in bulk tissue samples?

Accepted Answer

Single-cell RNA-sequencing (scRNA-seq) can reveal cell type-specific eQTLs by integrating single-cell RNA-sequencing data with population genetics from the same cohort of donors. This methodological framework allows for the identification of eQTLs that are unobservable with bulk RNA-seq. By analyzing the gene expression signatures in cell populations, scRNA-seq provides a deeper understanding of the cell type-specific effects of genetic variation on gene expression. This approach helps to overcome the limitations of heterogeneous cellular compositions in bulk tissues and enables researchers to explore genetic regulation at higher resolutions and larger scales. The integration of scRNA-seq and genetic profiles from a large number of individuals allows for the evaluation of genetic regulatory effects in all cell types, providing valuable insights into the underlying biological mechanisms of complex traits and diseases.

Accepted Answer

The Huatuo framework analyzes genetic regulation by performing a genome-wide analysis on single-cell transcriptome profiles. It consists of four main stages: integrating sequence information using a CNN model, fitting gene expression with XGBoost regression models, estimating chromatin-level effects of in silico mutagenesis, and performing integration analysis with population-based associations. This framework aims to disentangle causality for genetic regulation and seeks statistical support for variant predictions from population-based associations.

Accepted Answer

Huatuo predictions show a significant correlation with eQTL summary statistics, with a median Spearman R of 0.30 across tested tissues. This correlation indicates that the predicted variant effects within a linkage disequilibrium (LD) block align with the standard eQTLs from the GTEx consortium. The strong correlation suggests that Huatuo can accurately reproduce the results of population-based association analysis based on DNA sequences, making it a reliable tool for variant prediction.

Accepted Answer

Cell cluster-ieQTLs are interaction expression quantitative trait loci that modify gene expression levels with a cell cluster-dependent manner. They are mapped to provide more insights into the genetic regulatory effects in bulk tissue samples. For instance, in arterial tissue samples from GTEx individuals with a higher G allele dosage, the expression level of ZC3H3 was negatively correlated with the estimated cell cluster proportion of B cells. The association between the SNP genotype and the gene expression was stronger in tissue samples with a higher estimated cell cluster proportion of B cells. Cell cluster-ieQTLs showed significantly more overlap with reported ieQTLs compared to standard eQTLs, revealing variant-gene associations not observed with standard eQTLs. The colocalization analysis revealed a large number of colocalized signals not observed with standard eQTLs, showing significant enrichment for the gene-trait in the 'silver standard' dataset. Cell cluster-ieQTLs can originate from tested cell types, but other cell types with proportion estimates related to the tested ones may also contribute. The de novo variant predictions in Huatuo provide a solution to characterize the cell type specificity of ieQTLs. Cell type-specific ieQTLs are defined as those whose genotype main effects maintain a consistent sign and show a decrease from low to high estimated cellular proportions. The replication analysis showed that, on average, 37% of ieQTL SNP-gene pairs for B cells, T cells, monocytes, and NK cells were replicated in their matched OneK1K cell types. The replication rates of matched cell types consistently higher than those of other ieQTL cell types. The cell cluster-ieQTLs were significantly enriched in the putative regulatory variants combined from the cell cluster predictive models, showing the comparability between cell cluster-ieQTLs and predictions from cell cluster models in the discovery of cell type-specific genetic regulatory effects.

Accepted Answer

The Huatuo framework analyzed a comprehensive landscape of cell type-specific genetic variation for 44 major cell types from 20 human tissues. This analysis identified 13,182 cell type-specific functional regulatory variants and 6181 associated genes likely to be perturbed by putative causal regulatory variants. The framework also benchmarked these variants against associated cell type-specific open chromatin regions from an external scATAC-seq dataset, showing overall consistency between inferred and measured cell type-specific regulatory regions.

Accepted Answer

Huatuo prioritizes cell types for complex traits and diseases by performing cell type-specific stratified LD score regression using the Huatuo genetic variation landscape. This method estimates the marginal increases in SNP-based heritability enrichment for each cell type. The Huatuo-based prioritization allows for the recapitulation of well-known mechanisms, such as identifying vascular smooth muscle cells and oligodendrocytes as top enriched cell types for stroke and schizophrenia, respectively. The analysis also confirms previous findings, like associations of lupus with B cells, asthma with granulocytes, and rheumatoid arthritis with T cells. Additionally, significant associations between immune-related phenotypes and structure cells, such as multiple sclerosis with endothelial cells, psoriasis with epithelial cells, and rheumatoid arthritis with fibroblasts, are observed. These findings support the use of Huatuo in facilitating novel insights into the associations between cell types and complex traits and diseases.

Accepted Answer

Huatuo's genetic variation landscape can provide systematic insights into the mechanisms of complex traits and diseases by mapping causal variants to the putative cell type-specific genetic regulation at single-nucleotide resolutions. SuSiE finemapping analysis was performed to acquire the 95% credible sets, which are the smallest variant set to have a causal variant driving the observed GWAS association signals. The identified functional regulatory variants were tested to confirm that the causal variants altering gene expression levels and affecting complex phenotypes were shared. Across 114 GWAS datasets, 699 putative cell type-specific regulatory variants were nominated, with 672 genes correlated with the variants affecting the biology relevant to the corresponding phenotypes. For example, rs10791824 represents a functional variant with a specific transcriptional-regulation-disrupting effect in granulocytes, showing a significant association with AP5B1 gene expression abundance in peripheral blood. This analysis can reflect cellular contribution to the genetic risk factors, as seen in immunerelated phenotypes, where related transcriptional regulation in immune cells underlies the majority of causal variants. Additionally, genetic regulation in structure cells contributes to a fraction of phenotype-causal genetic variation, consistent with cell type heritability enrichment.

Accepted Answer

Huatuo is a versatile framework that enables systematic exploration of human noncoding genetic variation at both cell type and single-nucleotide resolutions. It generates a comprehensive landscape of cell type-specific genetic variation across human tissues, assessing how inferred genetic regulation contributes to complex traits and diseases. Huatuo prioritizes phenotype-driver cell types and provides functional insights into genetic associations with diseases. It overcomes limitations of current technology by integrating standard eQTLs and cell cluster-ieQTLs, enabling high-resolution uncovering of novel genetic regulation of gene expression. Huatuo is a valuable resource for precision medicine and personalized healthcare.

Accepted Answer

The study compared the highest absolute values of variant predictions and the maximum z-scores of eQTLs within the same LD block to calculate correlation across all genes. It also tested if ab initio variant predictions could distinguish fine-mapped variants from linked negative variants. The control sets were constructed by randomly selecting linked variants with a posterior inclusion probability (PIP) less than 0.01. Logistic regression classifiers were used to determine if a variant was the fine-mapped SNP of eQTLs. This research provides insights into the effectiveness of ab initio variant predictions in reproducing standard eQTL results and distinguishing fine-mapped variants.

Accepted Answer

Ab initio variant predictions were used by training a cell type model with HUDEP 2 cells' expression profiles. In silico mutagenesis was performed around the HBG TSS, and 10,000 null sets of variants were constructed. The fold difference was calculated by comparing the mean absolute value of predicted effects for pathogenic variants to that of null variants. Confidence intervals were assessed through 10,000-time bootstrapping of the constructed null sets with replacement.

Accepted Answer

PP.H4 values represent the posterior probability of having the same causal variant between two association data. In colocalization analysis, a PP.H4 value greater than 0.85 is considered evidence of positive colocalization. This means that the same genetic variant is likely to be responsible for both the GWAS associations and the cell cluster-ieQTLs or standard eQTLs. By using this threshold, researchers can prioritize genes that show strong colocalization signals, indicating a potential causal relationship between the genetic variant and the trait of interest. The use of PP.H4 values helps validate the colocalization results and provides a quantitative measure to assess the strength of the colocalization evidence.

Accepted Answer

To compare cell cluster-ieQTLs with ab initio variant predictions, the smallest eigenMT-adjusted p value of top associations per gene is used to rank cell cluster-ieQTLs. The R package liger is then used to estimate enrichment of genetic loci linked to predicted functional variants. Enrichment p-values are calculated through 10,000 permutations. Additional tissue-level models are trained using expression profiles of bulk tissues to obtain variants with cell type-specific regulatory effects. The predicted functional variants are divided into two groups based on their predicted effects. The enrichment of cell cluster-ieQTLs is recalculated for each group, and the normalized enrichment score (NES) is calculated as the enrichment of the observed variant set divided by the mean enrichment of 10,000 sample permutations.

Accepted Answer

To validate cell type-specific functional variants, download chromatin accessibility profiles for 54 cell types, perform enrichment analysis on 42 matching cell types, calculate overlap between open chromatin regions and predicted functional variants, and estimate enrichment using AUCell_calcAUC. Normalize scores and cluster with Ward.D2 algorithm. This process validates the specific enrichment of functional variants in corresponding cell types' open chromatin regions.

Accepted Answer

Logistic regression classifiers are constructed using the training dataset of maximum absolute values of ab initio variant predictions and strongest summary statistics of cell cluster-ieQTLs and standard eQTLs across investigated cell types. These classifiers distinguish between positive association signals (GWAS p-values < 5e-8) and negative signals (p-value > 0.05). They are then used to analyze GWASs with more than 20 defined positive association signals, investigating the global contributions of cell type-specific genetic regulation to complex diseases and traits. Annotation input files for LDSC v.1.0.1 20 are prepared using identified cell type-specific functional variants, and SNP-based heritability enrichment is estimated for 114 different GWASs. Genotype reference from the Phase 3 of the 1000 Genomes Project 45 is used to estimate LD scores, and enrichment p-values are corrected using the Bonferroni adjustment. QQ plots are applied to examine the inflation of identified variants in summary statistics of several GWASs below the Bonferroni-corrected significance threshold, using the R package qqman v.0.1 46.

Accepted Answer

Cell type-specific stratified LD score regression is a method used to estimate the marginal increases in SNP-based heritability enrichment for each cell type. It involves jointly fitting annotations such as identified cell type-specific functional variants, predicted variants, and a baseline model with 52 functional categories. The analysis tests whether identified variants for a cell type show significant enrichment for per-SNP heritability, controlling for other annotations. Genotype reference from the 1000 Genomes Project Phase 3 is used for LD score estimates. The enrichment score is defined as the signed log10-transformed values of heritability enrichment multiplied by p-values.

Accepted Answer

The purpose of Huatuo-based variant-to-function mapping is to provide insights into the biological mechanisms of phenotypecausal genetic variation. It involves systematic mapping of causal variants identified by GWASs to the putative cell type-specific genetic regulation from Huatuo at single-nucleotide resolutions. This mapping helps in understanding the potential intermediate phenotypes that cause phenotype variances, when Huatuo identified regulatory variants are in the estimated 95% credible sets of disease-causal and trait-causal SNPs. The process involves fine mapping analysis using SuSiE 28 and generating correlation matrices using PLINK v.1.9. The results are visualized using circos plots, providing a comprehensive understanding of the genetic regulation and its impact on phenotypes.

Accepted Answer

Research design details can be found in the Nature Portfolio Reporting Summary linked to the article. This summary provides comprehensive information about the research design, methodology, and data collection process. It is an essential resource for researchers and readers to understand the context and approach of the study. The Reporting Summary is available in an Open Access format, allowing anyone to access and utilize the information freely. It is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution, and reproduction in any medium or format. However, it is crucial to give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made. The article also includes images or other third-party material, which are covered by the same Creative Commons license unless otherwise specified. If the material is not included in the Creative Commons license and the intended use exceeds the permitted use, permission must be obtained directly from the copyright holder. To view the license, visit http://creativecommons.org/licenses/by/4.0/. This ensures that the research design details are accessible and can be utilized by other researchers for further studies or comparisons.

An analytical framework for decoding cell type-specific genetic variation of gene regulation

Chat with Paper

AI Agents for this Paper

Most frequently asked questions

1. How can single-cell RNA-sequencing (scRNA-seq) reveal cell type-specific eQTLs?

2. How does Huatuo framework analyze genetic regulation?

3. How do Huatuo predictions correlate with eQTL summary statistics?

4. What are cell cluster-ieQTLs and how do they provide insights into genetic regulatory effects in bulk tissue samples?

5. How many cell types were analyzed in the Huatuo framework?

6. How does Huatuo prioritize cell types for complex traits and diseases?

7. How can Huatuo's genetic variation landscape provide insights into complex traits?

8. What is Huatuo framework?

9. How do ab initio variant predictions compare to standard eQTLs?

10. How were ab initio variant predictions used to identify pathogenic variants?

11. What is the significance of PP.H4 values in colocalization analysis?

12. How are cell cluster-ieQTLs compared to ab initio variant predictions?

13. How to validate cell type-specific functional variants?

14. How are logistic regression classifiers used in identifying genetic regulation?

15. What is cell type-specific stratified LD score regression?

16. What is the purpose of Huatuo-based variant-to-function mapping?

17. Where can research design details be found?

Citations

An analytical framework for decoding cell type-specific genetic variation of gene regulation

Mapping Cell Atlases at the Single-Cell Level.

Deep learning approaches for non-coding genetic variant effect prediction: current progress and future prospects

xQTLatlas: a comprehensive resource for human cellular-resolution multi-omics genetic regulatory landscape

Variant Impact Predictor database (VIPdb), version 2: trends from three decades of genetic variant impact predictors

References

Gene set enrichment analysis: A knowledge-based approach for interpreting genome-wide expression profiles

PLINK: A Tool Set for Whole-Genome Association and Population-Based Linkage Analyses

XGBoost: A Scalable Tree Boosting System

A global reference for human genetic variation.

XGBoost: A Scalable Tree Boosting System

Related Papers (5)

Linkage Disequilibrium Patterns and Functional Analysis of RGS4 Polymorphisms in Relation to Schizophrenia

Characterization of Single Nucleotide Polymorphisms in 55 Disease-Associated Genes in a Korean Population

Common deletions and SNPs are in linkage disequilibrium in the human genome.

In Search of Human Variation

Remapping the type I diabetes association of the CTLA4 locus.