Deconvolution of complex molecular signals of metabolomics and transcriptomics for predicting drug sensitivity and therapy resistance in malignant melanoma

Tumour and TE heterogeneity are reflected in multimodal data including e.g. genomic, transcriptomic and metabolic levels. However, data integration and sharing functional annotations between different data modalities, is still a challenging task. Recently we proposed [38, 39] a deconvolution-based method which allows to connect several modalities. This will be applied on transcriptomic and metabolic datasets from the consortium and public datasets for predicting drug sensitivity and associated biomarkers in melanoma.