Dragon 7.0

Dragon has been discontinued. If you currently own a Dragon license and need technical support, please contact us at chm@kode-solutions.net

Dragon is the world-wide most used application for the calculation of molecular descriptors. Its new version, Dragon 7.0, provides an improved user interface, new descriptors and additional features such as the calculation of fingerprints and the support for disconnected structures.

  • Molecular descriptors: Dragon calculates 5,270 molecular descriptors, covering most of the various theoretical approaches. The list of descriptors includes the simplest atom types, functional groups and fragment counts, topological and geometrical descriptors, three-dimensional descriptors, but also several properties estimation (such as logP) and drug-like and lead-like alerts (such as the Lipinski’s alert). The wide range of different approaches and theories for descriptors calculation, and the correctness and precision of their implementation are ensured by the scientific supervision of the Milano Chemometrics and QSAR Research Group of prof. Roberto Todeschini, author with dr. Viviana Consonni of Molecular Descriptors for Chemoinformatics (the most complete reference for descriptors theory)
  • User interface and batch calculation: Dragon provides an easy-to-use and intuitive graphical user interface (both on Windows and Linux platforms) and a command line interface, useful for the batch processing of large datasets.
  • Fingerprints and molecular fragments: Dragon now provides also the calculation of hashed molecular fingerprints, fully customizable with several parameters, along with the generation of all molecular fragments used in the fingerprinting procedure.
  • Analysis tools: the graphical user interface provides advanced tools to analyze the calculated descriptors (extended univariate statistics, pair-wise correlation, principal component analysis) and the possibility of importing user’s defined variables (like available experimental values) and perform analysis on the merged set of calculated descriptors and user’s variables.
  • Disconnected structures: starting from version 7.0, Dragon allows the calculation of descriptors also on molecules with disconnected structure (e.g. salts, ionic liquids), providing different theoretical approaches to extend descriptors algorithms on such structures.
Software Type
Standalone application for Windows and Linux platforms
Requirements
Supported platforms: Microsoft Windows (7, 8.1 and 10) and Linux (Fedora Core 23, Ubuntu 14.04)

Dragon 7.0 calculates 5,270 molecular descriptors, organized in different logical blocks as in the previous versions. Blocks are further divided into sub-blocks to make management, selection, and analysis of descriptors easier. Following, the summary of molecular descriptors blocks calculated by Dragon 7.0 is reported.

BLOCK NO
BLOCK NAME
DESCRIPTORS
1 Constitutional 47
2 Ring descriptors 32
3 Topological indices 75
4 Walk and path counts 46
5 Connectivity indices 37
6 Information indices 50
7 2D matrix-based descriptors 607
8 2D autocorrelations 213
9 Burden eigenvalues 96
10 P-VSA-like descriptors 55
11 ETA indices 23
12 Edge adjacency indices 324
13 Geometrical descriptors 38
14 3D matrix-based descriptors 99
15 3D autocorrelations 80
16 RDF descriptors 210
17 3D-MoRSE descriptors 224
18 WHIM descriptors 114
19 GETAWAY descriptors 273
20 Randic molecular profiles 41
21 Functional groups count 154
22 Atom-centred fragments 115
23 Atom-type E-state indices 172
24 CATS 2D 150
25 2D Atom Pairs 1596
26 3D Atom Pairs 36
27 Charge descriptors 15
28 Molecular properties 20
29 Drug-like indices 28
30 CATS 3D 300

Dragon 7.0 provides the calculation of hashed molecular fingerprints.

It calculates path based and extended connectivity fingerprints (ECFP). Path based fingerprints are generated identifying all linear paths of the molecule while ECFP are generated exhaustively identifying all circular fragments grown radially from each heavy atom of the molecule.

Both fingerprint types are calculated defining a set of options like the fingerprint size, the minimum and maximum length of the identified fragments and an extensive list of atom properties to be considered in order to differentiate fragments (e.g. atom type, aromaticity, formal charge, connectivity and bond order). The full customization of fingerprints allows to obtain the most suitable solutions for any task (modelling, similarity screening etc.).

Along with the fingerprints, Dragon 7.0 also calculates the full list of molecular fragments identified in the processed dataset, and provides them as SMARTS string, together with a set of useful statistics.

Dragon is used as a part of several QSAR modelling applications and suites, here you can find a list of some projects and applications that use Dragon for the calculation of molecular descriptors.

  • CzeekS is a computational molecule screening software. It utilizes a machine-learning method known as a Support Vector Machine (SVM) which has been shown to have efficient pattern recognition ability. Using existing assay data sets, CzeekS generates prediction models for compound-protein interactions. Based on the Chemical Genomics-based Virtual Screening (CGBVS) method, CzeekS offers fast and high-precision in-silico screening for drug candidate compounds. CzeekS employs approximately 1,000 molecular descriptors calculated by Dragon.
  • Toxicity Estimation Software Tool (TEST) estimates the toxicity values and physical properties of organic chemicals from molecular structure by means of several QSAR models. TEST is released by the United States Environmental Protection Agency (US EPA) and it is one of the QSAR tools suggested as an alternative approach for the REACH legislation. Dragon was used as benchmark software for the calculation of the molecular descriptors included in the TEST models.
  • CAESAR is an European project specifically dedicated to develop QSAR models for the REACH legislation. Five endpoints are addressed: BCF, Skin Sensitization, Carcinogenicity, Mutagenicity, Developmental Toxicity. A free on-line application has been developed to make models fully usable by everyone. Most of the models are based on Dragon descriptors.
  • ChemGPS-NP Web is a web-based public tool, for comprehensive chemical space navigation and exploration in terms of global mapping onto a consistent, eight dimensional map over structure derived physico-chemical characteristics. ChemGPS-NPWeb can assist in compound selection and prioritization; property description and interpretation; cluster analysis and neighbourhood mapping; as well as comparison and characterization of large compound datasets. By using ChemGPS-NPWeb, researchers can analyze and compare chemical libraries in a consistent manner. Through this web-based interface the users can submit a SMILES/SD file and receive the output as a result file showing the positions on the ChemGPS-NP map in terms of calculated predicted scores that then can be visualized. This works by calculating 40 selected Dragon descriptors used to define the ChemGPS-NP map from the submitted structure-file.
  • Chembench is a free portal that enables researchers to mine available chemical and biological data. Chembench can help researchers rationally design or select new compounds or compound libraries with significantly enhanced hit rates in screening experiments. Chembench provides cheminformatics research support to molecular modelers, medicinal chemists and quantitative biologists by integrating robust model builders, property and activity predictors, virtual libraries of available chemicals with predicted biological and drug-like properties, and special tools for chemical library design. Chembench was initially developed to support researchers in the Molecular Libraries Probe Production Centers Network (MLPCN) and the Chemical Synthesis Centers.
  • MOLE db – Molecular Descriptors Data Base is a free on-line database comprised of 1124 molecular descriptors calculated for 234773 molecules. The molecules are mainly collected from the NCI database, while the molecular descriptors have been calculated by means of DRAGON. Basically, the MOLE db – Molecular Descriptors Data Base allows the user to search for a specific group of molecules and analyse the corresponding values of molecular descriptors.
  • The Materials and Processes Simulations – MAPS platform developed by Scienomics, is a comprehensive environment for molecular modeling and simulations. MAPS provides a flexible framework for accessing the best simulation technology required for materials modeling and design in chemical, pharmaceutical and materials research projects. MAPS’ plug-in based architecture enables users to integrate any proprietary or third party code when needed. MAPS has functionality to build molecular systems of finite, crystalline and amorphous nature and enables the user to visualize them with high quality graphics. Highly efficient engines in the area of quantum, classical, mesoscale, correlative and thermodynamic modeling allow predictions of relevant properties in various application areas including chemistry, petrochemistry, home and personal care, defense, transportation, pharmaceuticals and electronics. Sophisticated analysis tools are available for visualization and processing of the simulated results.
  • OCHEM Database (Online Chemical Database With Modeling Environment) is an online database with modeling environment. You can submit experimental data or use the other users’ data to build predictive QSAR models for physical-chemical or biological properties. Dragon is used as internal engine to calculate descriptors.
  • Eco-PDS (Eco-friendly Product Design Support tool) is an open on-line application to promote the development of eco-friendly chemical mixture products with screening the mixture toxicity which can be caused among chemicals. Eco-PDS is a web-based hazard assessment tool of mixtures that integrates four predictive models for estimating the toxicity of mixtures: i) Concentration Addition (CA) model; ii) Independent Action (IA) model; iii) Partial Least Squares-based Integrated Addition Model (PLS-IAM); and iv) QSAR-based Two Stage Prediction (QSAR-TSP) model. The QSAR-TSP model is based on Dragon descriptors.