The --dms option, added in v2.2, provides customized analyses and outputs for DMS-MaP experiments. Unlike SHAPE reagents, which react relatively uniformly with all four RNA nucleotides, DMS preferentially reacts with A and C nucleotides. Traditionally, it has been thought that DMS only reacts with A and C, but we have shown that it also reacts in a structurally informative way with U and G. The --dms mode is optimized to measure DMS reactivities from all four bases and normalize them to a consistent scale, as one would expect from a SHAPE experiment. Under common scenarios, DMS-MaP experiments performed using optimizied protocols and the --dms mode (together named enhanced DMS-MaP, or eDMS-MaP) outperform SHAPE-MaP experiments at all four nucleotides.
When running with the --dms flag, ShapeMapper will generate normalized .dms files in place of .shape files, and will also generate specialized profiles.pdf and histograms.pdf plots. DMS-specific quality checks are also performed, ensuring that U and G modification rates are high enough to be useful.
Appropriate buffering conditions are required to support robust U and G modification. Optimized Bicine buffer conditions are described in Mustoe et al, PNAS 2019, and Mitchell et al, Nucleic Acids Research, 2023,.
ShapeMapper --dms will automatically check for sufficient modification at U and G. If modification rates at one or both nucleotides are too low (for example, because of insufficient buffering capacity), then ShapeMapper will throw a warning and mask out all nts from downstream analyses. In this scenario, other nucleotides types (particularly A and C) may still contain usable data and will be reported.
The pipeline is optimized for datasets collected using the MarathonRT enzyme, which we recommend. But --dms analysis also benefits datasets collected using TGIRT-III and SuperScript II.
DMS reacts with the A-N1, C-N3, U-N3, and G-N1 and G-N7. G-N7 modifications occur with much greater frequency that G-N1, but only G-N1 modification contain information on RNA secondary structure. Fortunately, G-N1 modifications are preferentially encoded and G->C and G->U mismatches during MaP-RT, whereas G-N7 are encoded as G->A mismatches. ShapeMapper --dms filters out G->A mismatches to permit highly specific measurement of G structure. ShapeMapper --dms also filters out indels at all positions (i.e. only mismatches are counted), which contain less information than mismatches.
DMS reactivities are normalized independently for each nucleotide type (A, C, G, U) to comparable scales of [0, ~2]. DMS reactivities have a steeper dependence on structure than SHAPE reactivities: Whereas for SHAPE > 0.85 is considered highly reactive, for DMS >0.5 is considered highly reactive and indicative of single-stranded nucleotide.
We recommend using the Fold or Partition module of RNAstructure
to perform structure modeling.
We have developed structure modeling parameters that are optimized for the normalized .dms reactivities generated by ShapeMapper --dms. These parameter files are distributed in... [TODO]
Please cite Mitchell et al, Nucleic Acids Research, 2023, for publications using the --dms option
Bicine buffering conditions were first described in Mustoe et al, PNAS 2019