Abstract
MicroRNAs (miRNAs) are an important class of small noncoding RNA molecules that serve as excellent biomarkers of various diseases. However, current miRNA biomarkers, including those comprised of multiple miRNAs, work at a single-miRNA level but not at a miRNA-set level, which is defined as a group of miRNAs sharing common biological characteristics. Given the rapidly accumulating miRNA omics data, we believe that the miRNA-set level analysis could be an important supplement to the single-miRNA level analysis. Therefore, we present sTAM (http://mir.rnanut.net/stam), a computational tool for single-sample miRNA-set enrichment analysis. Moreover, we demonstrate the utility of sTAM scores in discovering miRNA-set level biomarkers through two case studies. We conduct a pan-cancer analysis of the sTAM scores of the "tumor suppressor miRNA set" on 15 types of cancers from The Cancer Genome Atlas (TCGA) and 14 from Gene Expression Omnibus (GEO), results of which indicated a good performance in distinguishing cancers from controls. Moreover, we reveal that the sTAM scores of the "brain development miRNA set" can effectively predict cerebrovascular disorder (CVD). Finally, we believe that sTAM can be used to discover disease-related biomarkers at a miRNA-set level.