Discovery and re-discovery of trancriptionally co-regulated biological processes

Co-classifications of two processes to the same gene were generated wherever the model identified a similarity of the expression profile of the classified gene with the profiles of training example genes of two different biological processes. Thus, co-classification revealed transcriptional co-regulation of one gene with genes participating in two different biological processes during the fibroblast serum response. High frequencies of co-classifications of a pair of processes to one gene obtained during re-classification (Table 8) indicated that many training genes from this pair of processes display similar temporal expression profiles.

Table 8 Pairs of cellular processes with high frequency of co-classifications to the same gene

Pairs of processes	Genes with co-classif.	Genes with co-annot.	Genes with correct co-classif.
DNA metabolism - Cell cycle	25	12	12
Transcription - Oncogenesis	22	7	5
Cell organization and biogenesis - Cell cycle	18	16	13
Transcription - Cell proliferation	18	12	10
Transcription - Cell cycle	18	9	7
Cell motility - Defense (immune) response	16	13	11
DNA metabolism - Transcription	15	10	9
Oncogenesis - Cell proliferation	14	8	5
Defense (immune) response - Cell proliferation	13	11	7
Transcription - Intracellular signalling cascade	13	5	2
Cell motility - Cell adhesion	12	11	10
Protein metabolism and modification - Stress response	12	5	5
Cell motility - Cell proliferation	11	8	6
Protein metab and modif - Cell organization and biogenesis	11	6	6
Cell proliferation - Intracellular signalling cascade	11	5	4
Transcription - Lipid metabolism	11	1	1

Our model thus hypothesized that some processes are related via transcriptional co-regulation during the fibroblast serum response. Many such pairs of processes, e.g. DNA metabolism - cell cycle, cell organization and biogenesis - cell cycle, and cell motility - defense (immune) response were also frequently co-annotated. This indicated that these processes are also related through an involvement of proteins known to participate in both processes. Consequently, our model re-discovered several pairs of processes that are also linked by co-annotations. Additionally, the model discovered transcriptional co-regulation of pairs of processes which do not involve high numbers of genes known to participate in both processes, such as, for instance, transcription - intracellular signalling cascade, protein metabolism and modification - stress response and transcription - lipid metabolism.