Xiaoxuan Xia
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Previous Organizations / Institutes
The organizations / institutes from the scholar's other articles in the past ten years
Research Subjects
Xiaoxuan Xia has a total of 43 co-authors. Here are the top co-authors by number of shared publications:
- Maggie Haitian Wang
Center for Clinical Research and Biostatistics, The Jockey Club School of Public Health and Primary Care, The Chinese University of Hong Kong
7 shared publications - Benny Zee
Center for Clinical Research and Biostatistics, The Jockey Club School of Public Health and Primary Care, The Chinese University of Hong Kong
6 shared publications - Ka Chun Chong
JC School of Public Health and Primary Care, Chinese University of Hong Kong, Hong Kong, China
6 shared publications - William Ka Kei Wu
Department of Anaesthesia and Intensive Care and Peter Hung Pain Research Institute, The Chinese University of Hong Kong, Hong Kong SAR, China
4 shared publications Rui Sun
The 7th Affiliated Hospital of Sun Yat-Sen University
3 shared publicationsHua Tang
Affiliation Department of Genetics, Stanford University, California, United States of America
2 shared publications
Latest Publications
- Open Access
Modeling structural variations sequencing information to address missing heritability and enhance risk prediction
Missing heritability remains a significant challenge in genome-wide studies focused on single nucleotide polymorphisms (SNPs) when analyzing the genetic basis of complex traits. Structural variations (SVs), which span broader genomic regions and often exert larger functional impacts than SNPs, hold promise for capturing this heritability. Although graph-based pangenomes have advanced SV detection and association analysis, traditional SNP-centric frameworks typically overlook detailed SV sequence information, treating overlapping SVs as independent variants and leading to reduced statistical power. To overcome this limitation, we developed SVrefiner , an algorithm that aligns overlapping SVs and partitions them into non-overlapping refined SVs (rSVs) based on sequence congruence. By generating precise genotype matrices, SVrefiner enables more accurate association studies. When applied to human, tomato, and pig genomes, SVrefiner delineated 48,712 rSVs from 77,696 human SVs, 6,607 rSVs from 51,561 tomato SVs, and 5,237 rSVs from 142,784 pig SVs. Incorporating rSVs alongside SNPs, indels, and unrefined SVs enhanced heritability estimates by up to 24.47%. Further, refined analysis revealed that gene expression traits are often influenced by specific SV subregions rather than entire SVs. Notably, integration of rSVs in human datasets increased total eQTLs by 71.56%, with cis- and trans- eQTLs rising by 21.3% and 94.4%, respectively. Mean risk prediction accuracy across over 16,000 traits improved by as much as 16.8%. These findings deepen the understanding of complex trait heritability and demonstrate the utility of rSVs for genetic improvement and disease risk assessment.
Open AccessModeling structural variations sequencing information to address missing heritability and enhance risk prediction
Missing heritability remains a significant challenge in genome-wide studies focused on single nucleotide polymorphisms (SNPs) when analyzing the genetic basis of complex traits. Structural variations (SVs), which span broader genomic regions and often exert larger functional impacts than SNPs, hold promise for capturing this heritability. Although graph-based pangenomes have advanced SV detection and association analysis, traditional SNP-centric frameworks typically overlook detailed SV sequence information, treating overlapping SVs as independent variants and leading to reduced statistical power. To overcome this limitation, we developed SVrefiner , an algorithm that aligns overlapping SVs and partitions them into non-overlapping refined SVs (rSVs) based on sequence congruence. By generating precise genotype matrices, SVrefiner enables more accurate association studies. When applied to human, tomato, and pig genomes, SVrefiner delineated 48,712 rSVs from 77,696 human SVs, 6,607 rSVs from 51,561 tomato SVs, and 5,237 rSVs from 142,784 pig SVs. Incorporating rSVs alongside SNPs, indels, and unrefined SVs enhanced heritability estimates by up to 24.47%. Further, refined analysis revealed that gene expression traits are often influenced by specific SV subregions rather than entire SVs. Notably, integration of rSVs in human datasets increased total eQTLs by 71.56%, with cis- and trans- eQTLs rising by 21.3% and 94.4%, respectively. Mean risk prediction accuracy across over 16,000 traits improved by as much as 16.8%. These findings deepen the understanding of complex trait heritability and demonstrate the utility of rSVs for genetic improvement and disease risk assessment.
- Open Access
- 27 October 2022
- Published by Public Library of Science (PLoS) in PLOS Genetics
- Vol. 18 (10) , e1010443
A Prism Vote method for individualized risk prediction of traits in genotype data of Multi-population
Multi-population cohorts offer unprecedented opportunities for profiling disease risk in large samples, however, heterogeneous risk effects underlying complex traits across populations make integrative prediction challenging. In this study, we propose a novel Bayesian probability framework, the Prism Vote (PV), to construct risk predictions in heterogeneous genetic data. The PV views the trait of an individual as a composite risk from subpopulations, in which stratum-specific predictors can be formed in data of more homogeneous genetic structure. Since each individual is described by a composition of subpopulation memberships, the framework enables individualized risk characterization. Simulations demonstrated that the PV framework applied with alternative prediction methods significantly improved prediction accuracy in mixed and admixed populations. The advantage of PV enlarges as genetic heterogeneity and sample size increase. In two real genome-wide association data consists of multiple populations, we showed that the framework considerably enhanced prediction accuracy of the linear mixed model in five-group cross validations. The proposed method offers a new aspect to analyze individual’s disease risk and improve accuracy for predicting complex traits in genotype data.
- 27 October 2022
- Published by Public Library of Science (PLoS) in PLOS Genetics
- Vol. 18 (10) , e1010443
Open AccessA Prism Vote method for individualized risk prediction of traits in genotype data of Multi-population
Multi-population cohorts offer unprecedented opportunities for profiling disease risk in large samples, however, heterogeneous risk effects underlying complex traits across populations make integrative prediction challenging. In this study, we propose a novel Bayesian probability framework, the Prism Vote (PV), to construct risk predictions in heterogeneous genetic data. The PV views the trait of an individual as a composite risk from subpopulations, in which stratum-specific predictors can be formed in data of more homogeneous genetic structure. Since each individual is described by a composition of subpopulation memberships, the framework enables individualized risk characterization. Simulations demonstrated that the PV framework applied with alternative prediction methods significantly improved prediction accuracy in mixed and admixed populations. The advantage of PV enlarges as genetic heterogeneity and sample size increase. In two real genome-wide association data consists of multiple populations, we showed that the framework considerably enhanced prediction accuracy of the linear mixed model in five-group cross validations. The proposed method offers a new aspect to analyze individual’s disease risk and improve accuracy for predicting complex traits in genotype data.
- Open Access
A Prism Vote Framework for Individualized Risk Prediction of Traits in Genome-wide Sequencing Data of Multiple Populations
Multi-population cohorts offer unprecedented opportunities for profiling disease risk in large samples, however, heterogeneous risk effects underlying complex traits across populations make integrative prediction challenging. In this study, we propose a novel Bayesian probability framework, the Prism Vote (PV), to construct risk predictions in heterogeneous genetic data. The PV views the trait of an individual as a composite risk from subpopulations, in which stratum-specific predictors can be formed in data of more homogeneous genetic structure. Since each individual is represented by a composition of subpopulation memberships, the framework enables individualized risk characterization. Simulations demonstrated that the PV framework applied with alternative prediction methods significantly improved prediction accuracy in mixed and admixed populations. The advantage of PV enlarges as the sample size, genetic heterogeneity, and population diversity increase. In two real genome-wide association data consists of multiple populations, we showed that the framework enhanced prediction accuracy of the linear mixed model by up to 12.1% in five-group cross validations. The proposed framework offers a new aspect to analyze individual’s disease risk and improve accuracy for predicting complex traits in genome data.
Open AccessA Prism Vote Framework for Individualized Risk Prediction of Traits in Genome-wide Sequencing Data of Multiple Populations
Multi-population cohorts offer unprecedented opportunities for profiling disease risk in large samples, however, heterogeneous risk effects underlying complex traits across populations make integrative prediction challenging. In this study, we propose a novel Bayesian probability framework, the Prism Vote (PV), to construct risk predictions in heterogeneous genetic data. The PV views the trait of an individual as a composite risk from subpopulations, in which stratum-specific predictors can be formed in data of more homogeneous genetic structure. Since each individual is represented by a composition of subpopulation memberships, the framework enables individualized risk characterization. Simulations demonstrated that the PV framework applied with alternative prediction methods significantly improved prediction accuracy in mixed and admixed populations. The advantage of PV enlarges as the sample size, genetic heterogeneity, and population diversity increase. In two real genome-wide association data consists of multiple populations, we showed that the framework enhanced prediction accuracy of the linear mixed model by up to 12.1% in five-group cross validations. The proposed framework offers a new aspect to analyze individual’s disease risk and improve accuracy for predicting complex traits in genome data.
- Open Access
- 16 July 2020
- Published by Springer Nature in Microbiome
- Vol. 8 (1) , 1-13
Bacteria pathogens drive host colonic epithelial cell promoter hypermethylation of tumor suppressor genes in colorectal cancer
Background: Altered microbiome composition and aberrant promoter hypermethylation of tumor suppressor genes (TSGs) are two important hallmarks of colorectal cancer (CRC). Here we performed concurrent 16S rRNA gene sequencing and methyl-CpG binding domain-based capture sequencing in 33 tissue biopsies (5 normal colonic mucosa tissues, 4 pairs of adenoma and adenoma-adjacent tissues, and 10 pairs of CRC and CRC-adjacent tissues) to identify significant associations between TSG promoter hypermethylation and CRC-associated bacteria, followed by functional validation of the methylation-associated bacteria. Results: Fusobacterium nucleatum and Hungatella hathewayi were identified as the top two methylation-regulating bacteria. Targeted analysis on bona fide TSGs revealed that H. hathewayi and Streptococcus spp. significantly correlated with CDX2 and MLH1 promoter hypermethylation, respectively. Mechanistic validation with cell-line and animal models revealed that F. nucleatum and H. hathewayi upregulated DNA methyltransferase. H. hathewayi inoculation also promoted colonic epithelial cell proliferation in germ-free and conventional mice. Conclusion: Our integrative analysis revealed previously unknown epigenetic regulation of TSGs in host cells through inducing DNA methyltransferase by F. nucleatum and H. hathewayi, and established the latter as CRC-promoting bacteria.
- 16 July 2020
- Published by Springer Nature in Microbiome
- Vol. 8 (1) , 1-13
Open AccessBacteria pathogens drive host colonic epithelial cell promoter hypermethylation of tumor suppressor genes in colorectal cancer
Background: Altered microbiome composition and aberrant promoter hypermethylation of tumor suppressor genes (TSGs) are two important hallmarks of colorectal cancer (CRC). Here we performed concurrent 16S rRNA gene sequencing and methyl-CpG binding domain-based capture sequencing in 33 tissue biopsies (5 normal colonic mucosa tissues, 4 pairs of adenoma and adenoma-adjacent tissues, and 10 pairs of CRC and CRC-adjacent tissues) to identify significant associations between TSG promoter hypermethylation and CRC-associated bacteria, followed by functional validation of the methylation-associated bacteria. Results: Fusobacterium nucleatum and Hungatella hathewayi were identified as the top two methylation-regulating bacteria. Targeted analysis on bona fide TSGs revealed that H. hathewayi and Streptococcus spp. significantly correlated with CDX2 and MLH1 promoter hypermethylation, respectively. Mechanistic validation with cell-line and animal models revealed that F. nucleatum and H. hathewayi upregulated DNA methyltransferase. H. hathewayi inoculation also promoted colonic epithelial cell proliferation in germ-free and conventional mice. Conclusion: Our integrative analysis revealed previously unknown epigenetic regulation of TSGs in host cells through inducing DNA methyltransferase by F. nucleatum and H. hathewayi, and established the latter as CRC-promoting bacteria.
- Open Access
- 24 December 2019
- Published by Springer Nature in BMC Medical Genomics
- Vol. 12 (S9) , 1-6
wtest: an integrated R package for genetic epistasis testing
Background: With the increasing amount of high-throughput genomic sequencing data, there is a growing demand for a robust and flexible tool to perform interaction analysis. The identification of SNP-SNP, SNP-CpG, and higher order interactions helps explain the genetic etiology of human diseases, yet genome-wide analysis for interactions has been very challenging, due to the computational burden and a lack of statistical power in most datasets. Results: The wtest R package performs association testing for main effects, pairwise and high order interactions in genome-wide association study data, and cis-regulation of SNP and CpG sites in genome-wide and epigenome-wide data. The software includes a number of post-test diagnostic and analysis functions and offers an integrated toolset for genetic epistasis testing. Conclusions: The wtest is an efficient and powerful statistical tool for integrated genetic epistasis testing. The package is available in CRAN: https://CRAN.R-project.org/package=wtest.
- 24 December 2019
- Published by Springer Nature in BMC Medical Genomics
- Vol. 12 (S9) , 1-6
Open Accesswtest: an integrated R package for genetic epistasis testing
Background: With the increasing amount of high-throughput genomic sequencing data, there is a growing demand for a robust and flexible tool to perform interaction analysis. The identification of SNP-SNP, SNP-CpG, and higher order interactions helps explain the genetic etiology of human diseases, yet genome-wide analysis for interactions has been very challenging, due to the computational burden and a lack of statistical power in most datasets. Results: The wtest R package performs association testing for main effects, pairwise and high order interactions in genome-wide association study data, and cis-regulation of SNP and CpG sites in genome-wide and epigenome-wide data. The software includes a number of post-test diagnostic and analysis functions and offers an integrated toolset for genetic epistasis testing. Conclusions: The wtest is an efficient and powerful statistical tool for integrated genetic epistasis testing. The package is available in CRAN: https://CRAN.R-project.org/package=wtest.
- Open Access
- 17 September 2018
- Published by Springer Nature in BMC Genetics
- Vol. 19 (S1) , 78
Incorporating methylation genome information improves prediction accuracy for drug treatment responses
An accumulation of evidence has revealed the important role of epigenetic factors in explaining the etiopathogenesis of human diseases. Several empirical studies have successfully incorporated methylation data into models for disease prediction. However, it is still a challenge to integrate different types of omics data into prediction models, and the contribution of methylation information to prediction remains to be fully clarified. A stratified drug-response prediction model was built based on an artificial neural network to predict the change in the circulating triglyceride level after fenofibrate intervention. Associated single-nucleotide polymorphisms (SNPs), methylation of selected cytosine-phosphate-guanine (CpG) sites, age, sex, and smoking status, were included as predictors. The model with selected SNPs achieved a mean 5-fold cross-validation prediction error rate of 43.65%. After adding methylation information into the model, the error rate dropped to 41.92%. The combination of significant SNPs, CpG sites, age, sex, and smoking status, achieved the lowest prediction error rate of 41.54%. Compared to using SNP data only, adding methylation data in prediction models slightly improved the error rate; further prediction error reduction is achieved by a combination of genome, methylation genome, and environmental factors.
- 17 September 2018
- Published by Springer Nature in BMC Genetics
- Vol. 19 (S1) , 78
Open AccessIncorporating methylation genome information improves prediction accuracy for drug treatment responses
An accumulation of evidence has revealed the important role of epigenetic factors in explaining the etiopathogenesis of human diseases. Several empirical studies have successfully incorporated methylation data into models for disease prediction. However, it is still a challenge to integrate different types of omics data into prediction models, and the contribution of methylation information to prediction remains to be fully clarified. A stratified drug-response prediction model was built based on an artificial neural network to predict the change in the circulating triglyceride level after fenofibrate intervention. Associated single-nucleotide polymorphisms (SNPs), methylation of selected cytosine-phosphate-guanine (CpG) sites, age, sex, and smoking status, were included as predictors. The model with selected SNPs achieved a mean 5-fold cross-validation prediction error rate of 43.65%. After adding methylation information into the model, the error rate dropped to 41.92%. The combination of significant SNPs, CpG sites, age, sex, and smoking status, achieved the lowest prediction error rate of 41.54%. Compared to using SNP data only, adding methylation data in prediction models slightly improved the error rate; further prediction error reduction is achieved by a combination of genome, methylation genome, and environmental factors.
- Open Access
- 17 September 2018
- Published by Springer Nature in BMC Proceedings
- Vol. 12 (S9) , 53
Gene-methylation epistatic analyses via the W-test identifies enriched signals of neuronal genes in patients undergoing lipid-control treatment
An increasing number of studies are focused on the epigenetic regulation of DNA to affect gene expression without modifications to the DNA sequence. Methylation plays an important role in shaping disease traits; however, previous studies were mainly experiment, based, resulting in few reports that measured gene–methylation interaction effects via statistical means. In this study, we applied the data set adaptive W-test to measure gene–methylation interactions. Performance was evaluated by the ability to detect a given set of causal markers in the data set obtained from the GAW20. Results from simulation data analyses showed that the W-test was able to detect most markers. The method was also applied to chromosome 11 of the experimental data set and identified clusters of genes with neuronal and retinal functions, including MPPED2I, GUCY2E, NAV2, and ZBTB16. Genes from the TRIM family were also identified; these genes are potentially related to the regulation of triglyceride levels. Our results suggest that the W-test could be an efficient and effective method to detect gene–methylation interactions. Furthermore, the identified genes suggest an interesting relationship between lipid levels and the etiology of neurological disorders.
- 17 September 2018
- Published by Springer Nature in BMC Proceedings
- Vol. 12 (S9) , 53
Open AccessGene-methylation epistatic analyses via the W-test identifies enriched signals of neuronal genes in patients undergoing lipid-control treatment
An increasing number of studies are focused on the epigenetic regulation of DNA to affect gene expression without modifications to the DNA sequence. Methylation plays an important role in shaping disease traits; however, previous studies were mainly experiment, based, resulting in few reports that measured gene–methylation interaction effects via statistical means. In this study, we applied the data set adaptive W-test to measure gene–methylation interactions. Performance was evaluated by the ability to detect a given set of causal markers in the data set obtained from the GAW20. Results from simulation data analyses showed that the W-test was able to detect most markers. The method was also applied to chromosome 11 of the experimental data set and identified clusters of genes with neuronal and retinal functions, including MPPED2I, GUCY2E, NAV2, and ZBTB16. Genes from the TRIM family were also identified; these genes are potentially related to the regulation of triglyceride levels. Our results suggest that the W-test could be an efficient and effective method to detect gene–methylation interactions. Furthermore, the identified genes suggest an interesting relationship between lipid levels and the etiology of neurological disorders.
- Open Access
- 17 April 2017
- Published by Wiley in Human Mutation
- Vol. 38 (9) , 1235-1239
Stratified polygenic risk prediction model with application to CAGI bipolar disorder sequencing data
Genetic data consists of a wide range of marker types, including common, low‐frequency, and rare variants. Multiple genetic markers and their interactions play central roles in the heritability of complex disease. In this study, we propose an algorithm that uses a stratified variable selection design by genetic architectures and interaction effects, achieved by a dataset‐adaptive W‐test. The polygenic sets in all strata were integrated to form a classification rule. The algorithm was applied to the Critical Assessment of Genome Interpretation 4 bipolar challenge sequencing data. The prediction accuracy was 60% using genetic markers on an independent test set. We found that epistasis among common genetic variants contributed most substantially to prediction precision. However, the sample size was not large enough to draw conclusions for the lack of predictability of low‐frequency variants and their epistasis.
- 17 April 2017
- Published by Wiley in Human Mutation
- Vol. 38 (9) , 1235-1239
Open AccessStratified polygenic risk prediction model with application to CAGI bipolar disorder sequencing data
Genetic data consists of a wide range of marker types, including common, low‐frequency, and rare variants. Multiple genetic markers and their interactions play central roles in the heritability of complex disease. In this study, we propose an algorithm that uses a stratified variable selection design by genetic architectures and interaction effects, achieved by a dataset‐adaptive W‐test. The polygenic sets in all strata were integrated to form a classification rule. The algorithm was applied to the Critical Assessment of Genome Interpretation 4 bipolar challenge sequencing data. The prediction accuracy was 60% using genetic markers on an independent test set. We found that epistasis among common genetic variants contributed most substantially to prediction precision. However, the sample size was not large enough to draw conclusions for the lack of predictability of low‐frequency variants and their epistasis.