A new algorithm has been developed for bioinformatics search of putative serum markers of cancer, which includes: 1) identification of microRNAs that are most often and most significantly overexpressed in tumors; 2) selection of mRNA targets regulated by microRNAs; 3) identification of mRNA targets encoding secreted proteins; 4) comparative analysis of mRNA transcription levels in normal and tumor tissues. Application of the algorithm led to discovery of seven putative serum markers of colon cancer: ADAMTS14, ANGPT2, CCL7, DEFA5, MMP11, MMP14, and PLAU. Experiments demonstrated that production of two out of seven proteins (MMP14 and DEFA5) is significantly increased in colon tumors vs. normal samples.