reactIDR: evaluation of the statistical reproducibility of high-throughput structural analyses towards a robust RNA structure prediction

Comprehensive in virio structure probing analysis of the influenza A virus identifies functional RNA structures involved in viral genome replication

The influenza A virus genome is segmented into eight viral RNAs (vRNA). Secondary structures of vRNA are known to be involved in the viral proliferation process. Comprehensive vRNA structures in vitro, in virio, and in cellulo have been analyzed. However, the resolution of the structure map can be improved by comparative analysis and statistical modeling. Construction of a more high-resolution and reliable RNA structure map can identify uncharacterized functional structure motifs on vRNA in virion. Here, we establish the global map of the vRNA secondary structure in virion using the combination of dimethyl sulfate (DMS)-seq and selective 2'-hydroxyl acylation analyzed by primer extension (SHAPE)-seq with a robust statistical analysis. Our high-resolution analysis identified a stem-loop structure at nucleotide positions 39 - 60 of segment 6 and further validated the structure at nucleotide positions 87 - 130 of segment 5 that was previously predicted to form a pseudoknot structure in silico. Notably, when the cells were infected with recombinant viruses which possess the mutations to disrupt the structure, the replication and packaging of the viral genome were drastically decreased. Our results provide comprehensive and high-resolution information on the influenza A virus genome structures in virion and evidence that the functional RNA structure motifs on the influenza A virus genome are associated with appropriate replication and packaging of the viral genome.

RNA Structure Determination by High-Throughput Structural Analysis

RNA functions are closely linked with their structures. Therefore, elucidating the secondary structure of RNAs provides crucial information regarding their function. The chemical modification or RNase-mediated digestion of single-stranded RNA has been utilized to experimentally reveal RNA secondary structures. Owing to advances in high-throughput sequencing technology and chemical analysis, RNA structural analyses that enable structural profiling at the transcriptomic scale in living cells have been developed. Here, we provide an overview of the high-throughput RNA structural (HTS) analyses and describe the computational processing steps of recent HTS analysis pipelines: PROBer, BUMHMM, and reactIDR.

Improving RNA secondary structure prediction via state inference with deep recurrent neural networks

The problem of determining which nucleotides of an RNA sequence are paired or unpaired in the secondary structure of an RNA, which we call RNA state inference, can be studied by different machine learning techniques. Successful state inference of RNA sequences can be used to generate auxiliary information for data-directed RNA secondary structure prediction. Typical tools for state inference, such as hidden Markov models, exhibit poor performance in RNA state inference, owing in part to their inability to recognize nonlocal dependencies. Bidirectional long short-term memory (LSTM) neural networks have emerged as a powerful tool that can model global nonlinear sequence dependencies and have achieved state-of-the-art performances on many different classification problems.

This paper presents a practical approach to RNA secondary structure inference centered around a deep learning method for state inference. State predictions from a deep bidirectional LSTM are used to generate synthetic SHAPE data that can be incorporated into RNA secondary structure prediction via the Nearest Neighbor Thermodynamic Model (NNTM). This method produces predicted secondary structures for a diverse test set of 16S ribosomal RNA that are, on average, 25 percentage points more accurate than undirected MFE structures. Accuracy is highly dependent on the success of our state inference method, and investigating the global features of our state predictions reveals that accuracy of both our state inference and structure inference methods are highly dependent on the similarity of pairing patterns of the sequence to the training dataset. Availability of a large training dataset is critical to the success of this approach. Code available at https://github.com/dwillmott/rna-state-inf.