
Xiaoyu Song, DrPH
Assistant Professor
Department of Population Health Science and Policy
Icahn School of Medicine at Mount Sinai
https://www.mountsinai.org/profiles/xiaoyu-song
Title: Cell-type-specific transcriptome wide association study of breast cancer risk
Abstract: Human mammary tissue is composed of multiple cell types with diverse roles in breast cancer etiology. Prior transcriptome-wide association studies (TWAS) of breast cancer have utilized methods that predict gene expression at the bulk tissue level without considering cell-type composition heterogeneity. Here we propose MiXcan, a novel extension of PrediXcan that predicts cell-type specific gene expression levels, identifies disease-associated genes and provides insights into the responsible cell type(s) of the associations. Compared with PrediXcan, MiXcan predicts gene expression levels with higher accuracy in the presence of cell-type heterogeneity and maintains comparable performance in the absence of cell-specific effects. An application to breast cancer study identifies 82 genes in MiXcan and only 31 genes in PrediXcan associated with breast cancer risk at a 10% false discovery rate. Genes that are uniquely identified by MiXcan are more likely to associate with breast cancer risk in epithelial cells, or show association with breast cancer risk in epithelial and non-epithelial cells in opposing directions. These findings suggest that cell-type-specific TWAS may reveal new insights into the genetic and cellular etiology of breast cancer and other diseases.
Dial-In Information
Please email Jiebiao Wang (jbwang at pitt.edu) for Zoom info.
Thursday, September 9 at 3:30 p.m. to 4:30 p.m.
Virtual EventXiaoyu Song, DrPH
Assistant Professor
Department of Population Health Science and Policy
Icahn School of Medicine at Mount Sinai
https://www.mountsinai.org/profiles/xiaoyu-song
Title: Cell-type-specific transcriptome wide association study of breast cancer risk
Abstract: Human mammary tissue is composed of multiple cell types with diverse roles in breast cancer etiology. Prior transcriptome-wide association studies (TWAS) of breast cancer have utilized methods that predict gene expression at the bulk tissue level without considering cell-type composition heterogeneity. Here we propose MiXcan, a novel extension of PrediXcan that predicts cell-type specific gene expression levels, identifies disease-associated genes and provides insights into the responsible cell type(s) of the associations. Compared with PrediXcan, MiXcan predicts gene expression levels with higher accuracy in the presence of cell-type heterogeneity and maintains comparable performance in the absence of cell-specific effects. An application to breast cancer study identifies 82 genes in MiXcan and only 31 genes in PrediXcan associated with breast cancer risk at a 10% false discovery rate. Genes that are uniquely identified by MiXcan are more likely to associate with breast cancer risk in epithelial cells, or show association with breast cancer risk in epithelial and non-epithelial cells in opposing directions. These findings suggest that cell-type-specific TWAS may reveal new insights into the genetic and cellular etiology of breast cancer and other diseases.
Dial-In Information
Please email Jiebiao Wang (jbwang at pitt.edu) for Zoom info.
Thursday, September 9 at 3:30 p.m. to 4:30 p.m.
Virtual Event