Cheminformatics & (HTS) QSAR

MOE provides a suite of applications for manipulating and analyzing large collections of compounds, building property models, consensus models and SD pipeline command line tools.

Operate directly on SD files for structure depiction, acid/base titration and tautomer enumeration, database filtering, sorting and descriptor calculations. Remove records that do not satisfy a series of filters (eg. lead-like, reactive groups, drug-like, etc.), sort records and remove duplicate entries from SD files. Calculate descriptors and write the output to SD or ASCII formats.

Use a unified small molecule tautomer and titration enumerator to prepare input structures for calculations or pharmacophore searching. The rule based method has conservative rules for strong acids and bases and borderline cases are enumerated. The application can be accessed from the graphical interface or the sdwash pipeline command tool.

Calculate over 600 molecular descriptors including topological indices, structural keys, E-state indices, physical properties (such as LogP, molecular weight and molar refractivity), topological polar surface area (TPSA) and CCG's VSA descriptors with wide applicability to both biological activity and ADME property prediction. Use descriptors for classification, clustering, filtering and predictive model construction. Add custom descriptors using MOE's built-in Scientific Vector Language.

PLS, PCR, Binary QSAR, & Recursive Partitioning Build QSAR/QSPR models using linear, probabilistic and decision-tree methodologies. CCG's unique Binary QSAR methodology is ideal for building pass/fail models from high error content data. Linear models include PCR and PLS methodologies and can support biological activity or ADME assessments.

Perform similarity searching and diverse subset selection using Descriptor, Conformation and Molecular Fingerprint methodologies. Choose between a number of fingerprint systems including 2, 3 and 4-point pharmacophore fingerprints in 2D or 3D and MACCS key fingerprints.

Design focused libraries using a product-based methodology that ranks individual reagents according to likelihood that they are part of an active compound. The ranking can be based on various types of models including linear and binary QSAR, fingerprint, pharmacophore and composite models. A Monte Carlo technique is used to avoid enumeration of the library allowing for reagent ranking in extremely large virtual libraries.

Use a Monte Carlo sampling technique to design large diverse combinatorial libraries. With this product-based methodology, full enumeration of the virtual library is avoided making it possible to extract diverse subsets when the chemistry space is extremely large.

Build moderately sized combinatorial libraries in MOE using a combinatorial library enumerator. Symmetric substitution, peptide substitution, bidentate connections and ring creation are supported (with appropriate treatment of chirality). Compounds are output to a MOE molecular database for subsequent visualization and analysis.

Quickly generate publication quality depictions of small molecules in 2D using a unique algorithm. The algorithm has been validated using a dataset of ~70,000 structures. Calculate 2D depictions for each molecule in a database and create depictions during ASCII import of SMILES strings.