Charting non-small cell lung cancer evolution and heterogeneity through multi-omics analysis and deconvolution

n many malignancies, molecular and cellular heterogeneity poses major challenges for diagnosis and treatment. Tumours are composed of subpopulations of cells harbouring distinct genomic alterations (subclones) which can affect their phenotype. Studies of intra-tumour heterogeneity generally focus on a single information layer. However, multi-omics characterisation is required to shed light on the functional impact of subclonal mutations. Multi-omics single cell sequencing of tumour samples is still technically challenging and costly when applied to larger cohorts. In this project, we will build on our bulk computational deconvolution methods, removing biases introduced by tumour purity and copy number, to reveal the pure tumour DNA methylation and gene expression profiles. Applied to the multi-region, multi-omics TRACERx dataset, we will explore DNA methylation and transcriptional evolution and heterogeneity in non-small cell lung cancer. Validation of our deconvolution will flow from data on sorted populations and small scale multi-omics single-cell sequencing. By directly mapping the deconvoluted epigenome and transcriptome onto established tumour phylogenetic trees, we will bridge clonal genotype and phenotype, revealing the interplay between the different omics layers and tumour clonal dynamics. This will aid our molecular understanding of cancer development and inform treatment strategies and biomarker design.