Abstract: The invention provides a method for identifying or detecting small RNA (sRNA) predictors of a disease or a condition. The method comprises identifying one or more sRNA sequences that are present in one or more samples of an experimental cohort, and which are not present across a comparator cohort; and optionally identifying one or more sRNA sequences that are present in one or more samples of a comparator cohort, and which are not present across an experimental cohort. In contrast to identifying dysregulated non-coding RNAs (such as miRs that are up- or down-regulated), the invention identifies sRNAs that are binary predictors, that is, present in one cohort (e.g., an experimental cohort) and not another (e.g., a comparator cohort). Further, by quantifying reads for individual sequences (e.g., iso-miRs), without consolidating reads to annotated reference sequences, the invention unlocks the diagnostic utility of miRs and other sRNAs.