Skip to main content
Fig. 1 | BMC Biology

Fig. 1

From: Joint learning improves protein abundance prediction in cancers

Fig. 1

Overview of the algorithm design for predicting proteomic expression from transcriptomic data. The overall task of this study is to transform the red matrix, representing the transcriptomic level expression across different individuals, to the blue-gray matrix, representing the proteomic level expression (left). Three models are created to address this problem (right): (1) generic model, which captures the innate correlation between mRNA and protein level; (2) gene-specific model, which captures how multiple genes work in a network to control the protein level under investigation through random forest aggregation of multiple base learners; and (3) trans-tissue model, which captures the shared functional networks across cancer types

Back to article page