ARSMA-web motif clips (motif-db)

The motif corpus of ARSMA-web: the coordinates of every hairpin loop, bulge and internal loop found in the RNA-containing entries of the Protein Data Bank, clipped out with their closing base pairs. It follows the RNA CoSSMos motif definition (CoSSMos 2.0: two closing base pairs on each side of the loop).

This is the table to count motif occurrences from. One row of a <class>.parquet file is one motif occurrence. 285,165 occurrences in 44 classes. Other ARSMA-web datasets (annotations, contacts, density maps, thermodynamics) describe these occurrences, and their row counts differ from this one for the reasons given in their own cards.

Files

file	one row is	notes
<class>.parquet	one motif occurrence	columns stem, pdb_id, internal_id, text (the clip as PDB-format text)
clips.sqlite	one motif occurrence	the same clips in one SQLite file (compressed text), for random access
manifest.csv	one motif occurrence	stem, motif_type, pdb_id, and which file holds the clip
_nr/nr_4.0A.parquet	one non-redundant occurrence	11,766 clips from BGSU representative-set chains (release 4.58, 4.0 Å); nr_4.0A.json records the source
_provenance/	one removed clip	clips removed on 2026-09-30 because their coordinates lacked residues their sequence names, or carried one closing pair instead of two
bad_clips*.csv, short_clips.csv, _to_delete/	one rejected clip	build audit logs. Clips listed there are not in the corpus

Classes

hairpinN is a hairpin loop of N unpaired nucleotides (3–13). bulgeN is a bulge of N nucleotides (1–5). AxB is an internal loop with A nucleotides on one strand and B on the other (1x1 … 7x7). The 25 classes CoSSMos defines are hairpin3–7, bulge1–5 and the internal loops up to 5x5 (asymmetric up to 4x5); ARSMA-web adds the other 19.

The motif identifier (stem)

<PDB>_<SEQUENCE>_<chain><residue>_<model>, for example 28UH_GCGAAAGC_a2375_1.

• SEQUENCE includes the two closing base pairs on each side, so a hairpin4 sequence is 8 nucleotides (2 + 4 + 2) and the loop is its middle four. Two-strand motifs list the two strands one after the other.
• residue is the first loop residue.
• model is the model number. NMR entries contribute one row per model, so the same site in a 20-model ensemble appears 20 times. Use model 1 if you need one row per site.

The stem is the key every other ARSMA-web dataset joins on (as clip_stem, motif_file without its .PDB suffix, or example_pdb).

Redundancy

About 86% of occurrences come from ribosome entries, many of them repeated depositions of the same ribosome. Report frequencies over _nr/nr_4.0A.parquet as well as over the full table.

Citation

Built on the RNA CoSSMos database (Vanegas et al., Nucleic Acids Res. 2012; Richardson, Kirkpatrick & Znosko, Database 2020). Coordinates are from the Protein Data Bank.