[open-science] Fwd: Silicos goes Open Source

Wed Jul 21 13:18:12 UTC 2010

---------- Forwarded message ----------
From: Wilfried Langenaeker <wilfried.langenaeker at silicos.com>
Date: Wed, Jul 21, 2010 at 12:49 AM
Subject: Silicos goes Open Source
To: "Bryan Bishop" <kanzure at gmail.com>

PRESS RELEASE:
SILICOS NV PORTS ITS PROPRIETARY COMPUTATIONAL CHEMISTRY SOFTWARE INTO THE
OPEN SOURCE DOMAIN

On June 22, 2010, the Belgium-based computational chemistry company
Silicos NV has made a
strategic decision to port the majority of its proprietary software
into the open source arena. The
decision has been made to port all of these tools and the
corresponding C/C++ API's into the Open
Babel environment under a GNU GPL licensing scheme. This strategic
decision will position
Silicos NV as one of the leading computational chemistry services
companies to support the open
source business model. According to Hans De Winter, Silicos' CSO, 'the
decision will allow Silicos to
move forward rapidly on the expanding wave of open source software
tools, and will significantly
expand its possibilities of providing services to customers in the
pharmaceutical and
biotechnological industry.'

About Silicos' proprietary software tools

Spectrophores™ are Silicos' patented 3D-field descriptors.
Spectrophores™ are a one-dimensional
description of three-dimensional molecular properties, such as atomic
electrostatic charges,
atomic lipophilicity, atomic softness, atomic electrophilicity and
shape. Spectrophores™ have been
succesfully used for clients to build sophisticated QSAR models in
combination with many
statistical modelling algorithms like bayesian classification and
support vector machines. Due to
their numeric nature, Spectophores™ are well-suited for the
calculation of self-organising maps.

Pharao is Silicos' implementation of a pharmacophore-based alignment
tool. It uses a Gaussian
representation of pharmacophoric features and generates an alignment
between pairs of a set of
pharmacophores. The resulting match is quantified using different
measures including a Tanimoto
coefficient.

Piramid is Silicos' software tool incorporating a shape-based
alignment between pairs of
molecules. Again, the resulting match is quantified using different
measures including a Tanimoto
coefficient.

Finally, Cosmos™ is Silicos' patented technology for the design of
novel molecules using an
evolutionary algorithm and using user-specified scoring functions. The
tool has been adapted to
be used in a fragment-based discovery enviroment. It has been
succesfully used in a number of
drug discovery programs in which Cosmos™ was able to generate novel
chemistry that has not
been described before. The strength of Cosmos™ lies in its
multi-objective optimisation algorithm
as well as in the fact that virtually all kinds of scoring functions
can be integrated.

About Silicos NV

Silicos NV is providing services in the field of computational
chemistry and virtual screening. The
company was founded in 2005 by two former employees of a major
pharmaceutical company in
Belgium and a former researcher at the ESAT-group of the Catholic
University of Louvain. Silicos is
a spin-out of the Catholic University of Louvain.

For more information

For more information regarding Silicos NV, please visit the website at
www.silicos.com or contact
Wilfried Langenaeker at +32 478 98 12 20.

Silicos NV
Wetenschapspark 7
B-3590 Diepenbeek
Belgium
www.silicos.com

"""
Silicos' virtual screening technology is based on a unique concept
called Spectrophores™. These are descriptors reflecting all possible
three-dimensional molecular properties which are involved in the
binding of a drug to its receptor, including electrostatic potential,
lipophilicity, softness, and shape.

Cosmos™ is Silicos' platform for the in silico, de novo design of
small molecules. The program serves as a fragment-based molecular
optimization application, in which novel molecules are generated by
combining fragments in a virtual synthesis step. These molecular
structures are further optimized in single or multi-objective
optimisation procedures. The evaluation of the quality of the
molecules can be ligand-based or target-based which makes Cosmos™ a
very flexible platform for de novo drug discovery.

Pharao is a pharmacophore-based tool to score and align small
molecules. The tool is based on the concept of modeling pharmacophoric
features by Gaussian 3D volumes instead of the more common point or
sphere representations. The smooth nature of these continuous
functions has a beneficent effect on the optimization problem
introduced during alignment.

SiMath is Silicos' open source API for statistical and mathematical
data modeling. SiMath is C++ library which integrates several open
source libraries into one library and which can easily included in
other applications.
"""

"""
De Novo design with Cosmos™

Cosmos™ is an acronym for Computational Optimization of Small
Molecular Structures. It is a highly flexible in silico platform,
based on a genetic algorithm for the evolution of molecules guided by
one or more evaluation functions and fragment-based virtual synthesis.
Evaluation functions can be ligand- and target-based.

As in a classical in vitro setup, Cosmos™ is build round three main
blocks: virtual synthesis, evaluation or virtual screening, and
optimization. The process is illustrated in the flowchart above. In
the virtual synthesis step, de novo molecules are developed from a set
of predefined building blocks and connectivity rules. Both building
blocks and connectivity rules have been extracted from known
compounds. As such, the possibility that the virtual compounds
actually are synthesized in a later stage is rather high. In the next
step, the molecules are evaluated using a virtual screening function.
The evaluation of the molecule can be based on any kind of function
and is completely flexible to meet the project needs. Typical
evaluators that have been tested, are docking, shape-based similarity,
fingerprint similarity, and Spectrophore™ models. The optimization
step can be either a single- or multi-objective optimization depending
on the number of targets and different scoring functions. From the
scores, molecules are selected for further participation in the
evolutionary process. Genetic operators, like mutation and crossover,
have been implemented again using the same building blocks and
connectivity rules as defined in the virtual synthesis. This closes
the loop and the process restarts.

Example

To illustrate the potential of Cosmos™, cisapride has been used as a
test case. The crystal structure of cisapride was used a template
structure and new virtually synthesised compounds were compared with
the template using a shape-based scoring function.
"""

- Bryan
http://heybryan.org/
1 512 203 0507