Grant: Discovery Grants Program - Individual
McGill University
$27,000.00
Statistical Modeling and Computational Methods for Spatial Genomic Data
FY 2023-24
Summary
The advanced technologies of imaging processing and spatially resolved genomics have enabled spatial data generation from different perspectives, which help to understand the physical organization of cells and to decipher biological mechanisms. The former technology processes histopathology images using numerous computational methods to generate biomedical object location data by performing abnormal tissue detection, cell identification, and cancer region segmentations. Meanwhile, the spatially resolved genomics technology can simultaneously profile gene expression and its spatial location. Despite the availability of spatial data generated from both perspectives, further statistical modelling is lacking to link the two sources of spatial information together and discover their statistical relationships. The long-term vision of this research program is to fill this gap by developing statistical methods to understand the relationship between high-dimensional spatial genomic data and the generated biomedical object location data. The short-term objectives of this research program address three challenges arising from the data: high dimension, measurement error, and sample correlation. The first objective is to build statistical models to quantify and test the spatial associations between high-dimensional gene expressions and the locations obtained from machine-learning algorithms. We will develop point process and Bayesian statistical methods to propose estimators and test statistics, which evaluate the association while incorporating the spatial features in the high-dimensional genomic data. The second objective is to address the measurement error or misclassification issue appearing in spatial data generation. The underline reasoning is that the spatial imaging biomarkers are often summarized using computational methods (e.g., deep learning), which are prone to errors or mistakes. Without adjusting for measurement error, the resulting estimators can be biased. It is essential to extend the models to incorporate measurement error in machine-generated data. The third objective focuses on the integration of the data from correlated samples. This is motivated by the availability of multiple slices collected from the same individual. This research program will have a significant impact on statistics and have a wide scope of applications in biological and medical research. The proposed models can enable practitioners to identify the gene set that might regulate the location of the cells. They can also be applied in oncology research to understand how genes influence the spread of cancer. Overall, this development of statistical methods and computational tools will be valuable for practitioners and can provide training for students in this newly emerging area.
Program
Discovery Grants Program - Individual
Recipient
Location
Quebec, Canada
Recipient Type
Academia
Type
Grant
Source
NSERC's Awards Data
As published
Application id
RGPIN-2023-03854
Name nom
Zhang, Qihuang
Department département
Epidemiology and Biostatistics
Organization id
61
Institution établissement
McGill University
Province en
Québec
Province fr
Québec
Country en
CANADA
Country fr
CANADA
Fiscal year exercice financier
2023
Competition year année de concours
2023
Award amount
27000
Program id
RGPIN
Program name en
Discovery Grants Program - Individual
Program name fr
Programme de subventions à la découverte - individuelles
Group en
DISCOVERY RESEARCH
Group fr
RECHERCHE AXÉE SUR LA DÉCOUVERTE
Committee code
1508
Committee name en
Mathematics and Statistics
Committee name fr
Mathématiques et statistique
Area of application code
1211
Area of application group en
Advancement of knowledge
Area of application group fr
Avancement des connaissances
Area of application en
Medical and health sciences
Area of application fr
Sciences médicales et de la santé
Research subject code
3001
Research subject group en
Statistics and probability
Research subject group fr
Statistical theory
Research subject en
Statistique et probabilité
Research subject fr
Théorie de statistique
Application title
Statistical Modeling and Computational Methods for Spatial Genomic Data
Keyword
association estimation; correlated data; image data; measurement error; misclassification; point process; spatial data; statistical genomics
Application summary
The advanced technologies of imaging processing and spatially resolved genomics have enabled spatial data generation from different perspectives, which help to understand the physical organization of cells and to decipher biological mechanisms. The former technology processes histopathology images using numerous computational methods to generate biomedical object location data by performing abnormal tissue detection, cell identification, and cancer region segmentations. Meanwhile, the spatially resolved genomics technology can simultaneously profile gene expression and its spatial location. Despite the availability of spatial data generated from both perspectives, further statistical modelling is lacking to link the two sources of spatial information together and discover their statistical relationships. The long-term vision of this research program is to fill this gap by developing statistical methods to understand the relationship between high-dimensional spatial genomic data and the generated biomedical object location data. The short-term objectives of this research program address three challenges arising from the data: high dimension, measurement error, and sample correlation. The first objective is to build statistical models to quantify and test the spatial associations between high-dimensional gene expressions and the locations obtained from machine-learning algorithms. We will develop point process and Bayesian statistical methods to propose estimators and test statistics, which evaluate the association while incorporating the spatial features in the high-dimensional genomic data. The second objective is to address the measurement error or misclassification issue appearing in spatial data generation. The underline reasoning is that the spatial imaging biomarkers are often summarized using computational methods (e.g., deep learning), which are prone to errors or mistakes. Without adjusting for measurement error, the resulting estimators can be biased. It is essential to extend the models to incorporate measurement error in machine-generated data. The third objective focuses on the integration of the data from correlated samples. This is motivated by the availability of multiple slices collected from the same individual. This research program will have a significant impact on statistics and have a wide scope of applications in biological and medical research. The proposed models can enable practitioners to identify the gene set that might regulate the location of the cells. They can also be applied in oncology research to understand how genes influence the spread of cancer. Overall, this development of statistical methods and computational tools will be valuable for practitioners and can provide training for students in this newly emerging area.