💻 bioinformatics

NCI-Profiler: automated protein-ligand interaction detection and visualization

The authors present NCI-Profiler, a free automated tool that detects and visualizes nine types of protein-ligand non-covalent interactions directly from structural coordinates with higher accuracy and speed than existing tools like PLIP, while offering enhanced coverage for complex cases such as metal coordination and macrocyclic ligands.

Original authors: Sharma, R., Lee, E. Y.

Published 2026-10-14

📖 5 min read🧠 Deep dive

CC BY 4.0

Original authors: Sharma, R., Lee, E. Y.

Original paper licensed under CC BY 4.0 (https://creativecommons.org/licenses/by/4.0/). ⚕️ This is an AI-generated explanation of a preprint that has not been peer-reviewed. It is not medical advice. Do not make health decisions based on this content. Read full disclaimer

In the microscopic world of biology, life depends on a constant, delicate conversation between molecules. Proteins, the workhorses of the cell, must recognize and bind to specific chemical partners to function, much like a lock accepting only the correct key. This binding is not a permanent fusion but a series of fleeting, invisible handshakes known as non-covalent interactions. These handshakes include hydrogen bonds, where atoms share a tiny electrical pull; salt bridges, which are stronger attractions between oppositely charged groups; and hydrophobic contacts, where water-repelling parts of molecules huddle together to escape the surrounding fluid. Scientists have long understood that the strength and specificity of these interactions determine how well a drug works or how an enzyme performs its job. However, as the library of known protein structures has exploded, manually inspecting these invisible handshakes has become impossible. Researchers need a way to automatically map these connections to understand why a drug binds to one target but not another, or to design new medicines that fit perfectly into the biological machinery.

To solve this problem, a team of researchers has developed a new digital tool called NCI-Profiler, designed to automatically find and draw these molecular connections. The tool acts as a universal translator for protein structures, taking raw data from a database and instantly identifying nine different types of interactions that hold a drug to its target. Unlike older programs that often miss subtle connections or struggle with complex, ring-shaped molecules, NCI-Profiler is built to catch a wider variety of interactions, including those involving metal ions and special bridges formed when a water molecule sits between a drug and a protein. The researchers tested the software against eighty-five different protein-drug complexes that had been carefully studied and described in scientific literature. The tool proved remarkably accurate, successfully identifying the vast majority of the interactions that human experts had previously confirmed. In many cases, it found connections that other popular software missed, and it did so roughly twice as fast.

The power of this tool lies in its ability to see what others overlook, particularly in the realm of halogenated drugs. Halogens are elements like chlorine, bromine, and iodine often added to medicines to improve their effectiveness. Sometimes, these atoms interact with proteins directly, but often they interact through a water molecule that acts as a bridge. Older tools frequently ignored these water-mediated bridges, treating the halogen and the protein as if they were not touching at all. NCI-Profiler, however, detects these three-way connections. In one real-world example, the researchers applied the tool to a fragment of a drug designed to fight a fungal infection. The software revealed that a chlorine atom on the fragment was forming a bridge with a water molecule, which in turn connected to a specific part of the protein's active site. This hidden connection pointed researchers toward a specific direction where the drug could be grown or expanded. Following this clue, they designed a new version of the drug that filled that space, displacing the water and binding nearly two hundred times more tightly than the original fragment.

The tool also helped explain why similar drugs work with different strengths. In a study of three versions of a drug used to treat erectile dysfunction, the researchers noticed that changing a single atom from chlorine to bromine to iodine changed the drug's potency. While the bromine version formed a direct, strong bond with the protein, the iodine version was even more potent despite failing to make a direct bond. NCI-Profiler showed that the iodine atom was instead forming a network of three water-mediated bridges, reaching out to touch multiple parts of the protein that the bromine version could not. This intricate web of connections explained the superior performance of the iodine drug, a detail that would have been invisible without the software's ability to map these water bridges.

Beyond finding new connections, the tool can also explain why a drug might work better on one type of receptor than another, a crucial factor in avoiding side effects. The researchers compared how a single drug molecule bound to two very similar receptors in the human body, one involved in sleep regulation and the other in mood. While the drug anchored itself to both receptors in the same way, the tool revealed a subtle difference in how the drug's ring-shaped structure interacted with a specific amino acid in each receptor. In the receptor where the drug bound more tightly, the interaction was a strong, flat stacking of rings, whereas in the weaker receptor, the rings were simply close but not perfectly aligned. This small geometric difference, detected automatically by the software, provided a clear structural reason for the difference in drug strength.

The researchers made this tool freely available to the public as both a website and a computer program that scientists can run on their own machines. It requires no special preparation of the data; users simply upload a file containing the atomic coordinates of a protein and its drug, and the system handles the rest. It produces clear, interactive diagrams that show the drug and its interacting neighbors, allowing scientists to rotate the view, hide specific connections, or download the image for reports. By automating the detection of these complex molecular relationships, NCI-Profiler offers a way to see the full picture of how drugs engage with the body, turning a fragmented and manual process into a streamlined, comprehensive analysis. This capability allows researchers to move faster from observing a structure to understanding the rules that govern it, potentially accelerating the discovery of safer and more effective medicines.

Technical Summary: NCI-Profiler

Problem Statement
Accurate detection and visualization of non-covalent interactions are fundamental to understanding protein–ligand binding specificity and affinity, yet current analysis tools present significant limitations. Existing profilers often cover a restricted set of interaction types, treat metal coordination merely as generic contacts, and struggle to represent macrocyclic ligands (e.g., heme, cobalamin) in automated 2D diagrams. Furthermore, critical interaction classes, such as halogen–water–hydrogen (XWH) bridges, are frequently omitted by standard web-based tools like PLIP and Arpeggio. Consequently, researchers often must fragment their workflows across multiple programs to achieve comprehensive interaction analysis, lacking a unified platform that integrates detection, editable 2D visualization, 3D structural context, and machine-readable output.

Methodology
The authors developed NCI-Profiler, an integrated web-based and command-line tool designed for the automated detection and visualization of nine distinct classes of protein–ligand non-covalent interactions. The system operates directly on PDB or mmCIF coordinates without requiring manual structure preparation, protonation, or format conversion.

Interaction Detection: The algorithm classifies molecular components (protein, solvent, ions, cofactors, ligands, glycans) using curated lookup tables validated against the wwPDB Chemical Component Dictionary. It evaluates nine interaction types—hydrogen bonds, salt bridges, halogen bonds, XWH bridges, metal coordination, $\pi$ – $\pi$ stacking, cation– $\pi$ interactions, hydrophobic contacts, and water bridges—using literature-derived geometric thresholds.
Specificity Cascade: To resolve redundancy where a single atom pair might satisfy multiple interaction criteria (e.g., a salt bridge also qualifying as a hydrogen bond), NCI-Profiler employs an eight-step specificity cascade. This prioritizes metal coordination, followed by salt bridges, halogen bonds, and hydrogen bonds, before applying ring-level and residue-level deduplication.
Specialized Handling:
- XWH Bridges: Explicitly identifies halogen–water–hydrogen bridges by requiring a simultaneous hydrogen bond between the bridging water and a protein residue, utilizing the $\sigma$ -hole model for halogen bond evaluation.
- Metal Coordination: Assesses a broad range of biologically relevant metal ions by comparing donor positions against reference geometries, computing $\tau_4$ and $\tau_5$ indices to distinguish coordination numbers.
- Macrocyclic Ligands: Handles complex macrocycles using a CCD-template-based strategy via PDBe CCDUtils, with a PCA-based coordinate-flattening fallback for ligands without dictionary entries.
Visualization and Output: The tool generates interactive 2D diagrams (SVG) with collision-resolved residue labels and a browser-based 3D viewer (Mol*). It also produces downloadable PyMOL sessions where ligands, residues, and interactions are organized as separate objects. Structured outputs are available in JSON, CSV, and plain-text formats.
Performance Optimization: Atomic charges for salt bridge detection are inferred via a geometry-based approach, eliminating the need for external protonation steps. This contributes to a reported speed advantage of approximately 2.2-fold over PLIP in benchmark tests.

Key Contributions

Unified Workflow: NCI-Profiler consolidates interaction detection, 2D/3D visualization, and structured data export into a single platform, supporting both individual structure analysis and batch processing via a Python CLI.
Expanded Interaction Coverage: It is the first tool to explicitly include XWH bridges as a distinct interaction class and to treat metal coordination as a specific geometric analysis rather than a generic contact.
Robust Handling of Complex Structures: The system automatically manages structures with multiple molecular components and macrocyclic ligands, reducing the need for manual curation.
Accessibility: The tool is freely available as a web server and a command-line interface (CLI) with a Python API, facilitating integration into computational pipelines and virtual screening workflows.

Results and Validation
NCI-Profiler was validated against a dataset of 85 literature-annotated protein–ligand complexes spanning all nine interaction types. Performance was measured against literature-reported interactions using precision, recall, and F1-scores.

Accuracy: The tool achieved F1-scores of 92–100% for $\pi$ – $\pi$ stacking, metal coordination, and water bridges. It scored 77–83% for cation– $\pi$ , XWH bridges, hydrogen bonds, salt bridges, and halogen bonds. Hydrophobic contacts yielded the lowest F1-score (54.8%), attributed to the lack of a universally accepted geometric definition in the literature rather than a detection failure.
Comparison with PLIP: On the same dataset, NCI-Profiler detected literature-confirmed interactions more completely (higher recall) in five of the nine categories compared to PLIP and ran approximately twice as fast.
Case Studies:
- Fragment-Based Drug Design (2UY3): Identified water-mediated halogen contacts (XWH bridges) at the periphery of a fragment, correctly predicting a growth vector that was later validated by a high-affinity analog.
- SAR Analysis (PDE5 Inhibitors): Resolved a shift from direct halogen bonding to water-mediated halogen contacts across a halogen series, explaining potency trends that conventional profilers missed.
- Subtype Selectivity (Melatonin Receptors): Distinguished binding modes between MT1 and MT2 receptors by detecting differences in $\pi$ – $\pi$ stacking geometry around a conserved phenylalanine, correlating with measured affinity differences.

Significance
The authors position NCI-Profiler as a specialized tool for scenarios where existing profilers have limited scope, particularly in the analysis of metalloenzyme inhibitors, halogenated ligands involving XWH bridges, and macrocyclic compounds. The paper claims that by broadening interaction-type coverage and enabling structured, direct comparisons across structures, NCI-Profiler provides genuine interpretive value for drug discovery tasks such as identifying binding-pocket growth vectors, explaining structure–activity relationships, and understanding receptor-subtype selectivity. The tool is intended to complement, rather than replace, established profilers, offering a distinct position in the landscape of structural biology tools through its interactive 3D visualization and comprehensive interaction profiling.

More like this