Best practices for genome-wide RNA structure analysis: combination of mutational profiles and drop-off information

Abstract

Genome-wide RNA structure maps have recently become available through the coupling of in vivo chemical probing reagents with next-generation sequencing. Initial analyses relied on the identification of truncated reverse transcription reads to identify the chemically modified nucleotides, but recent studies have shown that mutational signatures can also be used.While these two methods have been employed interchangeably, here we show that they actually provide complementary information. Consequently, analyses using exclusively one of the two methodologies may disregard a significant portion of the structural information. We find that the identity and sequence environment of the modified nucleotide greatly affects the odds of introducing a mismatch or causing reverse transcriptase drop-off. Finally, we identify specific mismatch signatures generated by dimethyl sulfate probing that can be used to remove false positives typically produced in RNA structurome analyses, and how these signatures vary depending on the reverse transcription enzyme used.

A mechanism for ligand gated strand displacement in ZTP riboswitch transcription regulation

“…To overcome this, we performed a separate experiment that enriched for transcript lengths 145 and 155 and used Superscript IV to improve full-length cDNA yield ( Figure 2A). Consistent with the observation that different RTs have distinct adduct detection biases, we find that SSIV reactivity measurements are sparse but in qualitative agreement with measurements made using SSIII (Novoa, Beaudoin, Giraldez, Mattick, & M., 2017;Sexton, Wang, Rutenberg-Schoenberg, & Simon, 2017) (Figure 2A). The reduced reactivity at G25, A31, and A40 observed with SSIII in the presence of ZMP after transcript length 117 is recapitulated across all post-termination site lengths with SSIV, suggesting that the ZMP-bound ON state persists beyond the point at which the terminator can form when the RNA is cotranscriptionally folded (Figure 2A).…”

Estimating RNA structure chemical probing reactivities from reverse transcriptase stops and mutations

“…Interestingly, this invariance is also suggested when observing the individual RT-stop and RT-mutate reactivities from Sexton et al 52 and Novoa et al 71 , which strongly suggest that adding the two together would create reactivities that are highly similar between RT enzymes and conditions.…”

“…However, recent papers 52, 71 have shown not only that these metrics are poorly correlated but also, in some contexts, completely orthogonal. Specifically, recent analysis of DMS probing RT-stop and RT-mutation signatures on the same pool of modified RNAs show that different reverse transcriptase enzymes and reaction conditions show distinct biases for detecting adducts as either RT-stops or RT-mutations 52,71 .…”

“…The major innovation in the above formula for chemical reactivities is to formally incorporate the observed signatures of both RT-stops and RT-mutations when estimating reactivities. The major motivation for this is the data and results of 52,72 which clearly show that RT drop and mutation signatures on the same pool of modified RNA differs depending on the specific RT enzyme and associated buffer conditions used to perform the conversion into cDNA. The goal of the RT process to convert every RNA adduct into a detectable signature in cDNAs therefore justifies our assertion that both RT-stops and RT-mutations should be included simultaneously in the reactivity estimation.…”

“…Accordingly, we expect two important improvements from applying the combined RT-stop+map model: improvement in chemical probing reactivity accuracy, and an invariance of reactivities to RT conditions. Improvements in chemical probing reactivity accuracy are naturally expected since current approaches that focus solely on RT-stops 3,11,18,30,34,36 or RT-mutations 32,46 will inherently miss information 52,72 . Many approaches to assess chemical probing accuracy rely on an indirect method to first utilize reactivities in RNA secondary structure prediction algorithms, and then assess accuracy of reactivity data based on the improvements in structural prediction [74][75][76][77] .…”