Molecular fingerprints, which efficiently describe molecular structures and allow rapid comparison of similarity, are particularly valuable in drug discovery. While 2D fingerprints are widely used, 3D structural interaction fingerprints offer a more detailed view of protein-ligand interactions and are especially useful in drug discovery projects. Flare™ facilitates the generation of such fingerprints, offering researchers an intuitive and efficient way to analyze complex molecular interactions with the new Protein-Ligand Interaction Fingerprints (PLIF) method. By leveraging PLIF, scientists can streamline workflows and gain deep insights into molecule-target interactions, benefiting projects such as:
- Insight into Structural Activity relationship (SAR) – interaction fingerprints provide insights into SAR by identifying similar interactions across a known ligand series with a defined binding site. This highlights key structural features that contribute to a molecule’s biological activity.
- Enhancing Virtual Screening Triaging – A virtual screen usually culminates in the selection of a diverse range of compounds which explore chemical space, usually based on structural diversity. Interaction fingerprint clustering can aid compound selection by also ensuring protein-ligand interaction diversity in the selection set, thereby keeping the number of compounds to be purchased and screened to a minimum.
- Fragment-Based Drug Discovery – Clustering of fragments based on their interactions with the protein binding site can identify ‘hot spots’ where multiple fragments bind (e.g. within an unoccupied binding site). This can subsequently guide the design of more potent molecules by combining structural features from different clusters.
- Efficiently exploring chemical space – Interaction clustering can efficiently identify diverse subsets of molecules from a chemical library, ensuring that a broad range of structural features are explored, highlighting under- or over-represented areas of a chemical space.
What Are Protein – Ligand Interaction Fingerprints?
Protein-ligand Interaction Fingerprints (PLIFs) in Flare can rapidly identify interaction motifs that are present in a molecular data set. The PLIFs are instrumental in clustering ligands, or fragments, based on their interactions with the target.
PLIFs comprise of a string of digits, where each digit represents a specific interaction type (such as Hydrogen bonds, Cation-Pi, Sulfur-lone pair, Halogen bonds, aromatic-aromatic interactions, salt-bridge-metal, hydrophobic interactions and steric clashes) on a per-residue basis. These strings (or fingerprints) can then be clustered using a hierarchical clustering method (Figure 1).
Figure 1. PLIFs for ligand 1 and 2. A count is created for every interaction type against a specific residue and added to a series of digits which forms the fingerprint. In this example, all interacting residues make one hydrogen bond, except the labeled methionine residue which forms two hydrogen bonds. Ligand 1 interacts with two fewer residues than Ligand 2 therefore resulting in two ‘0’ digits in its Fingerprint.
How to generate PLIFs in Flare?
PLIFs can easily be generated by selecting ‘PLIF Clustering’ in the ‘3D Pose’ tab (Figure 2) in a Flare project. This will open the ‘PLIF Clustering Calculation’ panel, where the ligands to cluster can be selected; whether the preferred pose or all poses of the ligands should be considered; and which protein-ligand contacts should be included in the PLIFs for clustering.
Figure 2. PLIF clustering example on the virtual hits obtained from a TNKS2 Blaze™ virtual screen.
The PLIF clustering analysis generates a new ‘PLIF Cluster View’ window that displays a dendrogram of the molecules (Figure 3). By right-clicking on the dendrogram, this result can be summarized in the Ligands table using Tags or a new column containing the cluster membership ID. The distance threshold can be adjusted (default=0.5; for Blaze virtual screening output a higher value might be preferrable (e.g. 0.8)).
Figure 3. A dendogram of the molecules is generated and displayed in the ‘PLIF Cluster view’ window which can easily be summarized in the ‘Cluster ID’ column or ‘Tags’ column in the Ligands table.
If the column data is chosen, two columns are generated: ‘PLIF Cluster ID’ and the ‘PLIF Cluster Size’, the latter providing the number of molecules in the cluster (Figure 4). PLIF gives the fingerprint information and can this be accessed by copy/pasting the [data] placeholders into a text editor or the Flare log file.
Figure 4. The ‘PLIF Cluster ID’ and ‘PLIF Cluster Size’ columns and ‘Cluster ID Tags’ have been generated to easily identify molecules from the same cluster that have similar interactions to the target protein.
Try PLIF Clustering on your project
PLIF clustering in Flare can rapidly identify compounds that have similar binding properties within a ligand data set. By clustering ligands based on their interaction profiles, PLIFs can enable efficient virtual screening triaging as well as provide critical insight into structure activity relationships. Whether guiding fragment-based discovery or exploring chemical space, this new feature offers significant potential to streamline and accelerating drug discovery projects.
If you would like to try the PLIF clustering feature on your project, request a free evaluation of Flare today.
