Quan Nguyen

Analysis of Spatial Data (Multiple student projects)

Nguyen group’s research is focused on understanding cancer complexity at tissue level by applying single-cell sequencing, spatial transcriptomics and tissue imaging, statistical learning and deep learning, and high performance computing. Most molecular biological data are from dissociated cells, which were separated from their original tissues, and thus the spatial connectin information is missing. Furthermore, these data often represent average measurements of millions of cells, which mask subtle differences that are specific for individual cells. From sequencing and imaging data, the group aims to computationally reconstruct biological regulatory networks underlying human diseases in every single cell within an indissociated tissue, like a tumour. The group develops both experimental and analytical methods to integrate genomics and imaging data for earlier and more accurate diagnosis and prognosis of diseases in tissue biopsies. Particularly, the group focuses on cancer (brain and skin cancer) and neuronal inflammation responses. Through advancing the understanding of biomarkers and cellular regulatory networks that are specific to individuals and cell types, the group contributes to early disease diagnosis, targeted drug discovery and precision medicine.

Traineeships, honours and PhD projects include

• Analyse spatial transcriptomics data of brain and skin cancer tissue to find cell-cell interactions, cell-type specific responses and cancer microenvironment evolution
• Develop experimental approaches to study spatial transcriptomics of human cancer cells in brain cancer xenograft models
• Develop experimental approaches to study formalin-fixed tissue sections for human skin cancer tissue sections
• Develop analysis methods to combine sequencing and imaging data from spatial transcriptomics experiments of skin cancer tissue sections
• Develop analysis methods to combine spatial transcriptomics, immuno-fluorescence images and histopathological images
• Find single cell gene regulatory networks in healthy and diseased cells from single cell and spatial datasets of human skin cancer samples

Regulatory Networks Determining Cell Types and Cell States

This project aims to use single-cell gene regulation networks to predict cell types and cell states in healthy and diseased tissues.

Through cell differentiation and division, a single fertilised egg gives rise to ~37.2 trillion cells with remarkable variation in forms and functions to make up the human body. A long-sought research goal over the past 150 years is to understand cell types and their properties and how they affect health and respond to environments. Conventional methods to assess cell type variability often rely on a small number of pre-characterised biomarkers and use population average measurements of millions of cells per sample, which is limited in resolution, accuracy, sensitivity, specificity, and comprehensiveness. Diverse cellular phenotypes encoded by the same genome are results from the differential regulation of large gene expression networks with about 22,000 genes. ‘Cell type’ and ‘cell state’ represent persistent and transient cellular properties, which can be defined by data-driven, network-based approaches. A systems-biology approach, which utilises advances in the computational analysis of big biological data and single-cell technologies, can be the key to decode the biological program in every cell type in the human body, thereby leading to better understanding and control of organismal phenotypes at the single-cell level.

The international Human Cell Atlas consortium (HCA) will release the first draft atlas comprising ~30-100 million cells for 15 organ systems in 1-2 years. Although at least 10 billion cells representing all tissues will be generated for the complete Atlas (Regev et al., 2017), the number is still marginal, accounting for 0.02% of the total 37 trillion cells in the body. Therefore, computational approaches are needed to recapitulate how the cells program the shared genome sequence in a human body to create astoundingly diverse forms and functions. From quantitative measurements of thousands of genes expressed in every cell, it is possible to reconstruct gene regulatory networks (GRN), the cellular programs. Regulatory ‘rules/patterns’ for molecular interactions are universally applicable in both population and single-cell data, and thus can be used to integrate datasets at single-cell and bulk-sample levels to infer GRN. This project will use gene expression regulatory networks to systematically quantify differences between cell types and cell states at single-cell resolution based. We will apply established analysis methods as well as develop new algorithms and software to integrate high-resolution scRNA-Seq data with large-scale population transcriptomics, genetics and epigenetics data to reconstruct gene regulatory networks. The ultimate aim is to predict the cell type and cell state of an unknown cell, by comparing the cell’s gene expression values to the largest single-cell regulatory network database generated in this project. The research would enable to predict cellular programs for thousands of cell types, which should contribute to the unprecedented ability to control and reprogram cells, to detect aberrant cells, and to understand how cells respond to the environment. Particularly, this project will contribute to studying cancer cell types and cell states at single-cell levels.

Spatial omics and machine learning to study heterogeneity and interaction between cells in primary tissues

This project aims at studying cell-cell and gene-gene regulatory networks in primary tissues by deep machine learning analysis of population, single-cell and spatial omics data.

Advances in genomics technologies enable data generation at an unprecedented speed, both in scale (hundreds of thousands of samples) and resolution (single cell). Machine learning in human genomics is an emerging field, which uses the power of statistics and high-performance computers in combination with biological knowledge to extract new information relevant to disease diagnosis and treatment.

Personalised and precision medicine require system genomics research to resolve variability at the cell, tissue and inter-individual level (e.g. different genetic background, age, exposure to environment). While big data integration of population genetics and single-cell omics studies can address variability between isolated cells and between individuals, a particularly important information dimension that is currently lacking is the heterogeneity in cell type composition and cell-cell interaction within the physiological context of a tissue. Such information is lost due to cell dissociation, a requirement for almost all molecular genomics assays.

We will contribute to research in personalised and precision medicine through deciphering the complex heterogeneity between cell types, tissues, and individuals by comprehensively integrating single-cell and population genetics with spatial transcriptomics, a novel type of information that is just beginning to be measured at a genome scale. Traditional machine learning and recent deep learning approaches for integrating multimodal genomics datatypes from bulk and single cells and image data will be applied. The systematic understanding of regulatory networks and biomarkers in a physiological context, which is specific to individuals and cell types will contribute to early disease diagnosis, targeted drug discovery and precision medicine. The research will generate an important understanding of variation in molecular networks inside individual cells and among neighbouring cells in specific microenvironments and among distant cell types involved in multi-organ communication, all of which underlie causal relationships between genotype and phenotype. The student will enjoy a conducive learning and research environment to develop a unique combination of multidisciplinary expertise in experimental biology, systems biology, biostatistics, and bioinformatics, and artificial intelligence.