An approach to handling and interpretation of ambiguous data in transcriptome and proteome comparisons

A major challenge towards a comprehensive analysis of biological systems is the integration of data from different omics sources and their interpretation at a functional level. Here we address this issue by analysing transcriptomic and proteomic datasets from mouse brain tissue at embryonic days 9.5 and 13.5. We observe a high concordance between transcripts and their corresponding proteins when they were compared at the level of expression ratios between embryonic stages. Absolute expression values show marginal correlation. We show in examples, that poor concordance between protein and transcript expression is in part explained by the fact, that single genes give rise to multiple transcripts and protein variants. The integration of transcriptomic and proteomic data therefore requires proper handling of such ambiguities. A closer inspection of such cases in our datasets suggests, that comparing gene expression at exon level instead of gene level could improve the comparability. To address the biological relevance of differences in expression profiles, literature-data mining and analysis of gene ontology terms are widely used. We show here, that this can be complemented by the inspection of physical properties of genes, transcripts, and proteins.