Circular RNA (
circRNA), a kind of special endogenous
RNA, has been shown to be implicated in crucial biological processes of multiple
cancers as a gene regulator. However, the functional roles of
circRNAs in
breast cancer (BC) remain to be poorly explored, and relatively incomplete knowledge of
circRNAs handles the identification and prediction of BC-related
circRNAs. Towards this end, we developed a systematic approach to identify
circRNA modules in the BC context through integrating
circRNA,
mRNA,
miRNA, and pathway data based on a non-negative matrix factorization (NMF) algorithm. Thirteen
circRNA modules were uncovered by our approach, containing 4164 nodes (80
circRNAs, 2703 genes, 63
miRNAs and 1318 pathways) and 67,959 edges in total. GO (Gene Ontology) function screening identified nine
circRNA functional modules with 44
circRNAs. Within them, 31
circRNAs in eight modules having direct relationships with known BC-related genes,
miRNAs or disease-related pathways were selected as BC candidate
circRNAs. Functional enrichment results showed that they were closely related with BC-associated pathways, such as 'KEGG (Kyoto Encyclopedia of Genes and Genomes) PATHWAYS IN
CANCER', 'REACTOME IMMUNE SYSTEM' and 'KEGG MAPK SIGNALING PATHWAY', 'KEGG P53 SIGNALING PATHWAY' or 'KEGG WNT SIGNALING PATHWAY', and could sever as potential
circRNA biomarkers in BC. Comparison results showed that our approach could identify more BC-related functional
circRNA modules in performance. In summary, we proposed a novel systematic approach dependent on the known disease information of
mRNA,
miRNA and pathway to identify BC-related
circRNA modules, which could help identify BC-related
circRNAs and benefits treatment and prognosis for BC patients.