Background, Novelty and Importance:
The advent of transcriptomics technologies has led to an abundance of large datasets. Bioinformatics has been successfully used to data mine such large data sets but struggles to integrate different types of datasets due to data heterogeneity, high-dimensionality, missing values, and data quality issues. For this machine learning is a useful tool to create workflows to bring data together to allow meaningful, robust, analysis. As new technologies emerge, new workflows are required. This project focuses on developing a machine learning workflow for integrating scRNAseq datasets with the new technology of single-cell resolution spatial transcriptomics.
Spatial transcriptomics gives spatial information for individual cells within a given tissue allowing cell types and tissue structure to be resolved. Current methods are rapidly evolving but are often limited to set probe sets rather than providing genome-wide coverage. In contrast, scRNAseq characterises genome-wide transcription levels within individual cells in a given sample but does not capture the spatial positions of cells. The ideal would be to integrate datasets generated from the two platforms. However, this is challenging as both methods can produce data that are sparse and noisy with different dimensionalities of data.
In this project, a pipeline will be created taking into account quality checks, normalisation of expression level generated from the two platforms, and projection of cell types and spatial distribution allowing complementary integration of scRNAseq and spatial transcriptomics.
For this project we will focus on complex spatial data generated by the King’s Spatial Biology Facility, to be integrated with publicly available scRNAseq datasets.
Aims and Objectives:
1. To evaluate current methods for integrating diverse transcriptomic data
2. To develop a workflow with machine learning methods for data integration
3. To link genetic and environmental factors to gene expression changes to determine the impact of the microbiome on chemosensation.

