RMDB

RDAT format specification

The RDAT file format is a hierarchical, tab-separated text format for RNA structural mapping data. The current revision is 0.34.

File structure

An RDAT file has three sections:

  1. General header — version, name, construct sequence, structure, offset, comments.
  2. Construct annotation — one or more lines starting with ANNOTATION_DATA:<index> describing each variant/replicate.
  3. Data — REACTIVITY:<index>, REACTIVITY_ERROR:<index>, optional TRACE:<index> and READS:<index> lines (tab-separated values, one per nucleotide position in SEQPOS).

General header keys

Key Meaning
RDAT_VERSION Format version (currently 0.34)
NAME Human-readable construct name
SEQUENCE RNA sequence (single-letter, uppercase main / lowercase flanking)
STRUCTURE Secondary structure (dot-bracket)
OFFSET Number subtracted from SEQPOS indices to give the user-facing numbering
SEQPOS List of positions for which reactivity is reported
COMMENT Free-form notes (repeated allowed)
ANNOTATION Construct-level annotations (key:value pairs, tab-separated; repeated allowed)

Common ANNOTATION keys

chemical: (e.g. MgCl2:10mM), temperature: (24C), modifier: (DMS, 1M7, CMCT, NMIA, …), processing: (pipeline steps), reverse_transcriptase:, experimentType: (StandardState, MutateAndMap, …).

Construct-level vs data-level annotations

Annotations on the ANNOTATION line apply to the whole construct. Annotations on ANNOTATION_DATA:<i> lines apply only to data row i (e.g. a single mutant in a mutate-and-map series). The data row’s annotations override the construct-level ones for that row.

Example

A minimal RDAT file:

RDAT_VERSION    0.34
NAME            16S rRNA Four-Way Junction
SEQUENCE        GGGAAACUGCCUGAUGGAGGGG...
STRUCTURE       (((..((...(((((((.(((....
OFFSET          125
COMMENT         Standard-state SHAPE in 10 mM MgCl2.
ANNOTATION      chemical:MgCl2:10mM   temperature:24C   modifier:1M7
ANNOTATION_DATA:1    sequence:GGGAAACUGCC...    name:wt    reads:12000
REACTIVITY:1         0.12    0.08    0.34    ...
REACTIVITY_ERROR:1   0.01    0.01    0.02    ...

Real RMDB entries usually contain dozens to thousands of ANNOTATION_DATA / REACTIVITY rows.

Reference implementation

The Python reference parser is rdat_kit (Apache 2.0). The RDATFile class loads, validates, and round-trips files of all versions back to 0.1.

For validating a file against this spec, see the validate page.