News

Pentacle

What is Pentacle?

Pentacle is an advanced software tool for computing, handling and interpreting GRid INdependent Descriptors (GRIND). It was born as a replacement of the pioneering ALMOND software. All our experience in the practical application of GRIND was applied to obtain a powerful, flexible and friendly tool.Pentacle introduces the next generation of GRIND (called GRIND-2) which solve some of the most important limitations of the original GRIND. In addition, the program has been completely rebuilt, implementing an advanced graphical interface that makes the application of GRIND easier than ever.
In addition, the program includes new and exciting features, like the integration of QSAR and Virtual Screening tools. For the nostalgic, the original GRIND were also implemented.

GRIND-2, the next generation of alignment-independent descriptors.

The GRIND were published in 2000 as the first MIF-based alignment-independent descriptors. Since its publication many applications have been published in different fields of drug design and development; QSAR, library design, ADME, virtual screening, chemogenomics, etc. At July 2008, the original work accumulated more than 130 citations. The use of GRIND in both academic and private research centers is now widespread.However, along the years some users have critisized several aspects of the original GRIND. Most of these problems have been solved in a new generation of GRIND that we called GRIND-2. Here are some examples:

In the original algorithm the success of the application often required to adjust a couple of parameters (number of nodes and field weights) by trial and error methods.

The algorithm used to simplify the MIF in the GRIND method included adjustable parameters. In addition, the results often contain mistakes, selecting too many nodes when no node should be obtained and too few for large compounds (see figures below). GRIND-2 solves the problems using a simpler algorithm called AMANDA, which requires no parameter, is much faster (up to 500 times faster) that the original one and provides a much more realistic representation of the relevant regions of the MIF.
You can compare the results of both methods in the following figure (new GRIND-2 in red, original GRIND original in purple).

The couples of nodes extracted for representing certain variables for rather similar compounds were not consistent, thus making harder the interpretation of the models.

The original GRIND variables represent couples of nodes located at a certain distance. When there are several couple located at that distance, the one with the highest field values is chosen. This simple criteria might lead to inconsistencies within series of related structures that makes the chemical interpretation more difficult. The GRIND-2 incorporate a novel method to make this selection with avoid inconsistences and, as a byproduct, produces a GRIND-based structural alignment of the compounds.

Pentacle is also an advanced Virtual Screening tool

Pentacle is able to process over 4K compounds/hour and CPU and take advantage of multi-core CPU architecture to obtain linear speedups. For example, in a dual QuadCore server Pentacle can process the whole DrugBank in less than 15 minutes.The GRIND are stored in a compact format which allows to obtain query results in a database with more than one million compounds in less than 20 seconds.

An advanced GUI

Pentacle is a powerful tool with a very intuitive GUI developed to be easy to use.All our expertise in using GRIND was applied in Pentacle development for obtaining an intuitive interface that presents the user with all the information needed at every working step (no more, no less). The whole GUI was built considering the separate task involved in the computation and handling of GRIND (structure importing, descriptor setup, descriptor inspection, model building and model interpretation). Every task was implemented in a separate tab of the main window which contains everything the user needs.
Unlike ALMOND, Pentacle does not need to open many separate windows and the access to the required information is always simple and fast. In particular, the GRIND QSAR models were easier to interpret than ever, thanks to a devoted interpretation wizard tool.
We have collected some screenshoots, so you can have an idea of the application “look and feel”

Please contact Molecular Discovery for further information about the software capabilities or a demo version.

Features and benefits

  • Simple and fast generation of GRIND and GRIND-2 descriptors
  • Automatic ionization of the structure to a given pH
  • Representation and handling of the GRIND using correlograms, heatmaps and 3D molecular graphics
  • Automatic GRIND-based structural alignment for easier model interpretation
  • Building QSAR models using PCA, PLS, and FFD variable selection
  • Fully integrated model interpretation tools (interpretation wizard)
  • Model library building and handling, including prediction of properties for new compounds
  • Virtual screening capabilities. Databases of over one million compounds can be easily built and interrogated
  • Windows, Linux32 and Linux64 versions

Technical specifications

  • Can operate both in GUI and command mode. Easily integrable in scripts.
  • Fully compatible Linux and Windows versions. Both can share databases, library model, computation templates and projects between different users or servers
  • Support for kout, mol2 and 3D sdf formats.
  • Modest system requirements. Even for virtual screening, a regulalar PC with 4Gb RAM can build and handle databases over one million compounds.

References

  • Pastor M, Cruciani C, McLay I, Pickket S, Clementi S. GRid-INdependent descriptors (GRIND): a novel class of alignment-independent three-dimensional molecular descriptors. J Med Chem. 2000 Aug 24;43(17):3233-43.
  • Durán A, Comesaña G, Pastor M. Development and validation of AMANDA, a new algorithm for selecting highly relevant regions in molecular interaction fields. J Chem Inf Model. 2008; 48: 1813-23. PMID: 18693718
  • Durán A, Zamora I, Pastor M. Suitability of GRIND-Based Principal Properties for the Description of Molecular Similarity and Ligand-Based Virtual Screening. J Chem Inf Mod. 2009; 49: 2129-38. PMID: 19728739

Pentacle has been developed by Ángel Durán and Manuel Pastor, with the financial support of Molecular Discovery. Commercial information can be found here.