Rapid Structure-Function Insights via Hairpin-Centric Analysis of Big RNA Structure Probing Datasets

Abstract

The functions of RNA are often tied to its structure, hence analyzing structure is of significant interest when studying cellular processes. Recently, large-scale structure probing (SP) studies have enabled assessment of global structure-function relationships via standard data summarizations or local folding. Here, we approach structure quantification from a hairpin-centric perspective where putative hairpins are identified in SP datasets and used as a means to capture local structural effects. This has the advantage of rapid processing of big (e.g., transcriptome-wide) data as RNA folding is circumvented, yet it captures more information than simple data summarizations. We reformulate a statistical learning algorithm we previously developed to significantly improve precision of hairpin detection, then introduce a novel nucleotide-wise measure, termed the hairpin-derived structure level (HDSL), which captures local structuredness by accounting for the presence of likely hairpin elements. Applying HDSL to data from recent studies recapitulates, strengthens, and expands on their findings which were obtained by more comprehensive folding algorithms, yet our analyses are orders of magnitude faster. These results demonstrate that hairpin detection is a promising avenue for global and rapid structure-function analysis, furthering our understanding of RNA biology and the principal features which drive biological insights from SP data.

Accurate Detection of RNA Stem-Loops in Structurome Data Reveals Widespread Association with Protein Binding Sites

“…Scores are further processed into c-scores via a normalization scheme based on an estimated distribution of scores associated with null sites (sites that do not harbor the target motif). For details on the core patteRNA algorithm, see (35) and (37); for details on score normalization, see (36).…”

“…The Scikit-learn Python module was utilized to perform these computations. Scripts that perform this quantification (and others, including the receiveroperating characteristic) are available in (37).…”

“…Typically, the user provides this motif in dot-bracket format, but patteRNA also has built-in routines to automatically mine some canonical motifs. For instance, patteRNA can automatically mine a representative set of hairpins (referred to as regular hairpins or regular stem-loops; defined as stem-loops with stem length between 4 and 15 nt, and loop length between 3 and 10 nt) via the "--hairpins" flag (37). When mining a particular structural element, only loci in the provided transcripts which satisfy the sequence constraints necessary for the target's secondary structure (via Watson-Crick and Wobble base pairs) are considered during scoring; these loci are henceforth referred to as "sites."…”

“…By scanning reactivity profiles alone, patteRNA was able to achieve reasonable accuracy when mining canonical motifs, such as hairpins/stem-loops (37). However, there was room for improvement via integration of NNTM-derived sequence information, which we believed could likely assist in situations where SP data is inconclusive.…”

“…Rather than perform complete RNA folding, it was developed to rapidly mine local structure elements from reactivity profiles via an unsupervised, versatile, and NNTM-free approach. In short, the method couples a statistical reactivity model-e.g., a Gaussian mixture model (GMM) or discretized observation model (DOM)-to a Hidden Markov model of structure (35)(36)(37). A parameterized model subsequently enables rapid quantitative scanning for locations that are likely to harbor a specific structure element.…”