ATLAS (Advanced Template Library for Assembly and Structure) is a comprehensive 3D RNA motif library
that enables researchers to search for RNA structural motifs from the Protein Data Bank. Using graph-based algorithms,
ATLAS catalogs over 433,000 RNA motif structures with atomic-level 3D coordinates.
What is ATLAS?
ATLAS features nucleotide-level graph representations of RNA structures incorporating both
Watson-Crick (WC) and non-Watson-Crick (non-WC) interactions.
The database includes five types of RNA motifs: hairpin loops, internal loops, bulges, multiway junctions,
and pseudoknots. Each motif includes 3D atomic coordinates retrieved from PDB files and graph representations
with and without non-WC interactions.
Motif Types in ATLAS
Hairpin Loops
Single-stranded loop closed by a stem
Internal Loops
Unpaired regions on both strands between base pairs
Bulges
Unpaired region on one strand only
Multiway Junctions
3-way to 10-way junction structures
Pseudoknots
Complex structures with crossing base pairs
How to Use ATLAS
1. Type-Based Motif Search
Search pre-indexed RNA motifs from the ATLAS database by motif type and size:
Step 1: Select Motif Type from the dropdown menu:
Hairpin - closed loops with unpaired nucleotides
Internal - unpaired nucleotides on both strands
Bulge - unpaired nucleotides on one strand
Multiway Junction - 3-way, 4-way, or higher-way junctions
Pseudoknot - H, HHH, Hlout, Hlin, LL, or LR types
Step 2: Select Motif Size (number of unpaired nucleotides) or junction type.
Note: The displayed size is the pure motif size; backend conversion handles stem nucleotides automatically.
Step 3: Click SEARCH to view results
Step 4: Download results:
CSV file - Contains motif metadata including PDB IDs, nucleotide numbers, and graph representations
ZIP file - Contains all matched PDB structures with atomic coordinates
2. Graph-Based Structure Search
Draw your own RNA motif structure using an interactive graph-based interface:
Step 1: Click Graph-Based Structure Search button on the homepage
Step 2: Build your motif structure:
Click Add Node to add nucleotides (nodes)
Connect nodes by selecting two nodes and clicking an edge type button
Edge Types:
Covalent Bond [0,0,1] - Backbone connections between adjacent nucleotides
Watson-Crick [1,0,0] - Standard WC base pairs (A-U, G-C)
Non-Watson-Crick [0,1,0] - Non-canonical base pairs (wobble, Hoogsteen, etc.)
Step 3: Use Load Example 1 to see a sample 4-node structure
Step 4: Click Submit Search - The system uses subgraph isomorphism algorithm
to find matching patterns (processing may take several minutes for complex structures; current timeout: 20 minutes)
Step 5: Download matched structures as CSV and PDB ZIP files from the results page
3. Understanding Search Results
PDB ID - Protein Data Bank identifier for the source structure
NT Number - Nucleotide residue numbers in the motif (comma-separated)
Motif Type - Classification of the structural motif
Filecontent - 3D atomic coordinates in PDB format
Graph Representation - Nucleotide-level graph with edge attributes for interaction types
Database Statistics
ATLAS contains over 433,000 unique motif entries extracted from RNA structures in the Protein Data Bank:
Internal loops: 51.8% (most common motif type)
Hairpin loops: 27.4%
Bulges: 13.2%
Junctions (3-way to 10-way): 6.0%
Pseudoknots: 1.7% (6 subtypes classified)
Technical Details
Graph Representation: Nucleotide-level graphs where nodes represent nucleotides and edges represent
interactions (covalent bonds, WC pairs, non-WC pairs) with one-hot encoding attributes
Search Algorithm: Subgraph isomorphism with NetworkX GraphMatcher for exact edge attribute matching
Data Source: RNA structures from PDB, processed with MC-Annotate for interaction analysis