In silico unwinding of microRNA duplexes to evaluate thermodynamic end stabilities improves predictions of microRNA strand selection.

microRNAs (miRNAs) are endogenous ∼22 nucleotide long, non-coding RNAs that post-transcriptionally regulate gene expression. During miRNA biogenesis, stem-loop-containing miRNA precursors are enzymatically cleaved to form a small RNA duplex. Cleavage positions are determined based on the position of structural motifs and junctions on the stem-loop precursor. The duplex end containing a favorable 5' nucleotide and lower thermodynamic stability is subsequently loading into an Argonaute protein. Typically, one duplex (guide) strand is retained in Argonaute and becomes functional whereas the other (passenger) strand is degraded. Therefore, accurate structural predictions of miRNA intermediates and quantification of duplex end stabilities are important towards understanding miRNA biogenesis. Here, we compiled predicted secondary structures for all miRNA hairpins and duplexes at physiologically relevant temperatures. We developed a new approach to calculate the thermodynamic stability of miRNA duplex ends, which resulted in improved predictions of miRNA strand selection. Our approach introduces hard constraints to folding algorithms to restrict base-paring of terminal nucleotides, which improves modeling of duplex end unwinding. We propose that constrained RNA folding can be used to evaluate local stabilities within an RNA secondary structure.