Use fPocketR Online
Commercially Available fPocketR No-Code Web Server
fpocketR: A Platform for Identifying and Analyzing Ligand-Binding Pockets in RNA
fpocketR is a software package that identifies, characterizes and visualizes ligand-binding pockets in RNA. It is built around fpocket, the open-source pocket detection software based on Voronoi tessellation and alpha spheres. The Weeks lab at UNC adapted it for RNA by optimizing its parameters against a curated set of RNA-ligand complexes.
fpocketR is ligand-agnostic. Docking tools need a candidate ligand and usually rely on a generic pocket finder to define the search space. fpocketR instead answers a different question: which RNAs, and which pockets within them, can bind drug-like ligands at all.
Why RNA Needs Its Own Pocket Finder
RNA is an enticing drug target, yet only a few human-devised small molecules have succeeded so far. The paper names linezolid, which binds the ribosome, and the splicing modifiers risdiplam and branaplam.
RNA is generally more polar, more flexible and more dynamic than protein. These differences make it hard to transfer protein-based tools and rules of thumb.
Only about 50 distinct RNA-ligand complexes with drug-like ligands have been visualized.
With default, protein-optimized parameters, fpocket detects many incorrect pockets in RNA. Its highest-scoring pocket overlaps the known ligand-binding site only 63% of the time.
Key Capabilities
Optimized for RNA: Four core fpocket parameters that control alpha sphere size and clustering were tuned for RNA.
Ligandability ranking: Pockets are ranked and described by their properties.
Visualization: Pockets are shown on both the 3D structure and the RNA secondary structure.
Ensemble analysis: Analyze every state of an NMR, cryo-EM or modeled ensemble to find transient pockets.
Flexible inputs: Provide a PDB accession code or a local RNA structure file.
How fpocketR Works
fpocketR runs the fpocket algorithm with RNA-optimized parameters, characterizes the pockets and any ligands, ranks pockets by ligandability, and writes a full set of outputs.
Inputs
A single input: a Protein Data Bank accession code or a locally stored RNA structure file
Optional RNA secondary structure drawing templates in NSD, VARNA or R2DT formats
Optional arguments to choose the RNA chain and ligand, adjust pocket-finding parameters, or analyze all states of an ensemble
Outputs
A 3D figure and PyMOL session file showing all pockets in the RNA tertiary structure
A detailed table of properties for each pocket and ligand
A secondary structure map showing which nucleotides form each pocket
For multi-state analysis, pocket density maps in both tertiary and secondary structure views
How It Was Optimized
Optimization used a curated training library of RNAs in complex with drug-like ligands. On a benchmark of 32 RNA-ligand complexes, the tuned version detected all known ligand-binding sites. It also raised the positive predictive value from 19% for standard fpocket to 78%.
What fpocketR Has Shown
Complex RNA Structure Makes Pockets
Multi-helix junctions and pseudoknots were found to be an order of magnitude more likely to form pockets for drug-like ligands than simpler motifs such as bulges and loops.
Large RNAs
fpocketR was trained mainly on short RNAs under 200 nucleotides, but it worked on three large cryo-EM structures: a 544-nucleotide RNA origami scaffold, a 374-nucleotide Pepper and Broccoli aptamer construct, and a 332-nucleotide fungal group I intron bound to an inhibitor. It identified every known pocket and seven novel ones. Four of the novel pockets formed at pseudoknots, and the others at a helix interface, a multi-helix junction and a G-quadruplex.
Ligand-Free (Apo) Structures
fpocketR detected the known ligand-binding site in 9 of 10 apo structures, including RNAs that change shape substantially on binding, such as the TPP riboswitch.
Computational Models
On 153 CASP15 models of the preQ1 riboswitch, 25% contained a pocket overlapping the known ligand site, 33% had only non-overlapping pockets, and 42% had no pocket. Model quality mattered. Among models with a TM-score above 0.26, 48% had a pocket at the known site. A 20-model SAM-IV riboswitch ensemble from trRosettaRNA put pockets in the same two regions as the cryo-EM ensemble.
Dynamic RNAs and Transient Pockets
Across six conformational states of the E. coli 70S ribosome during translocation, fpocketR found two transient pockets at the B2a inter-subunit bridge. They appear only in the INT2 and POST states and partially overlap binding sites of the antibiotics thermorubin, viomycin and capreomycin.
A Large Pocket Survey
On the Hariboss RNA-ligand library (365 complexes), fpocketR detected 139 known pockets, which overlapped ligands with higher drug-likeness scores, and identified 237 novel pockets.
Pocket Shape
Pockets detected by fpocketR are mostly flat, spanning rod-like to disc-like shapes. Their average shape closely matches that of FDA-approved drugs, which suggests RNA-specific shape is probably not the key property to focus on when building RNA-targeted libraries.
Experimental Validation
A companion fragment-based method called Frag-MaP detects ligand-binding sites in cells. The sites it found consistently surrounded pockets detected by fpocketR, supporting the ligandability of the novel pockets.
What is Tamarind Bio?
Tamarind Bio is a no-code bioinformatics platform built to give life scientists and researchers access to powerful computational tools. Many cutting-edge machine learning models are hard to deploy and use. Tamarind provides an intuitive, web-based environment that removes the complexity of high-performance computing, software dependencies and command-line interfaces.
The platform is designed for biologists, chemists and other researchers who may not have a background in programming or cloud infrastructure but want to run models on their own data. Key features include:
A user-friendly graphical interface for setting up and launching experiments
A robust API for integration into existing research pipelines
An automated system for managing and scaling computational resources
Tamarind treats information and data security as a top priority, as detailed in its Trust Center and Terms of Service.
Accelerating Discovery with fpocketR on Tamarind Bio
The authors point to several uses for RNA pocket detection:
Prioritize RNA targets: Find which RNAs, and which regions within them, can bind drug-like small molecules.
Discover new binding sites: The method flagged novel pockets in known structures that no ligand had been seen in.
Target dynamic RNAs: Identify conformation-specific pockets that could guide ligand design, such as the transient ribosome pockets.
Use modeled structures: When no experimental structure exists, RNA models can still indicate whether an RNA is ligandable, with better results from higher-quality models.
Guide antibiotic design: The authors suggest compact, drug-like ligands could be designed for the transient ribosome pockets.
Define docking search spaces: Use detected pockets to focus docking on realistic sites.
How to Use fpocketR on Tamarind Bio
Access the platform. Log in to tamarind.bio.
Choose your RNA structure. Use a PDB accession code or a local structure file. Higher-resolution structures give the best results, but models and ensembles also work.
Add a secondary structure template (optional). Provide an NSD, VARNA or R2DT drawing to get the pocket map in secondary structure space.
Run a single-state analysis. For one structure, use the open-source command-line tool from the fpocketR repository with the structure and template as inputs.
Analyze an ensemble. For NMR, cryo-EM, molecular dynamics or modeled ensembles, set the state option to 0 to analyze all states and map pocket density across them.
Refine the run. Optionally choose the RNA chain, name a bound ligand, set output resolution, and set the output folder. The authors also used a quality filter that dropped pockets scoring below 0.40.
Review the outputs. Open the PyMOL session and 3D figure, read the pocket and ligand property tables, and check the secondary structure map to see which nucleotides form each pocket.
Prioritize pockets. Use the ligandability ranking and pocket properties to choose sites for docking or experimental testing.
fpocketR needs a Unix-based operating system. On Windows, use WSL2 or a virtual machine.
Source: Veenbaas, Koehn, Irving, Lama & Weeks, Ligand-binding pockets in RNA, and where to find them, PNAS 122(17):e2422346122 (2025). DOI: 10.1073/pnas.2422346122. Companion paper: Veenbaas, Felder & Weeks, fpocketR: A platform for identification and analysis of ligand-binding pockets in RNA, bioRxiv (2025). DOI: 10.1101/2025.03.25.645323