Faerun

Faerun applies the 217-dimensional MXFP (macromolecule extended atom-pair fingerprint) to visualize and perform MXFP-based nearest-neighbor similarity searches of non-Lipinski PubChem (NLP) and non-Lipinski ChEMBL (NLC) molecules to characterize structural diversity and identify structurally similar macromolecular modalities.

Key Features:

• Dataset focus: Targets non-Lipinski PubChem (NLP) and non-Lipinski ChEMBL (NLC) subsets, including approximately seven million NLP entries in PubChem.
• MXFP fingerprint: Employs a specialized 217-dimensional macromolecule extended atom-pair fingerprint (MXFP) tailored for large molecules.
• Structural descriptors: MXFP captures molecular shape and pharmacophore-related features relevant to large and non‑small-molecule modalities.
• Visualization: Provides visualization of NLP and NLC molecular collections based on MXFP-derived similarity relationships.
• Similarity search: Supports MXFP nearest-neighbor searches to identify molecules with similar MXFP profiles within the NLP dataset.
• Problem addressed: Designed to address limitations of standard PubChem and ChEMBL search tools that are optimized for small molecules.

Scientific Applications:

• Structural diversity analysis: Characterizes the structural diversity of large, non-Lipinski molecules and novel modalities in PubChem and ChEMBL.
• Similarity-driven discovery: Identifies structurally similar macromolecules to support exploration of potential biological activities or therapeutic applications.
• Dataset interrogation: Enables focused analysis of large-molecule subsets (NLP and NLC) that are underrepresented by small-molecule–centric tools.

Methodology:

Uses the 217-dimensional MXFP (macromolecule extended atom-pair fingerprint) applied to NLP and NLC molecules for visualization and for MXFP nearest-neighbor similarity searches.