Transcript ICIC2007 DC Torus Patent

Dr. Matthew Wright
Product Director
www.digitalchemistry.co.uk
Torus
Molecule and Markush Structure Analysis within
Oracle
Digital Chemistry
• Company, formed Mid-2005
• Acquired Barnard Chemical Information (BCI)
September 2005.
– All previous BCI directors and staff are now
employed by Digital Chemistry.
• Aim to build on top of existing BCI technology
and know-how, predominantly their Markush
and Clustering methods
• Now moving into patent space
Digital Chemistry
• Product range covering
–
–
–
–
Cluster/diversity analysis
Markush library analysis
Chemical fingerprinting and dictionaries
Structure query conversion (MDL > Daylight)
• Software distributed as
– Toolkits (C, C++, Java, Perl, Python)
– Web Services
– Application suites
Torus overview
• Torus:Server
– Oracle based cartridge which integrates both Molecule and
Markush structure storage and retrieval
– SQL based interface, but Query can be supplied from
SMARTS, ISIS/Draw or ChemDraw
– Client may be Torus:View or any SQL supporting application,
e.g. PipeLine Pilot or customer developed system due to open
nature of SQL interface
• Torus:View
– Optional client for Torus:Server. Allows searching of
Torus:Server database with graphical display of results.
Markush Structures
• Presently within Torus:Server Markush structures have attachments
defined in a number of ways (with unlimited members), including :
Simple substitution :
R1
R1=
Nested R-Groups :
R1
Br
R1=
R2=
Bridged R-Groups :
Cl
*
*
O
N
R2
*
R2
*
Cl
Br
*
*
R1
R1 + R2 =
R2
O
*'
N
* ''
*'
* ''
Torus: Server Function summary #1
Torus:Server supports a number of SQL based functions:
• Structural equivalence check
– Are two libraries/molecules equivalent
• Sub-structure/Exact-structure searching
– Into Mol and Markush
• Overlap between Markush libraries
– Also produce third Markush representing overlap
• Similarity
– Of molecule into Markush and between two Markush libraries
• Property calculation, both in molecule and Markush
– Supports all Lipinski rules, e.g. weight, LogP etc
Torus: Server Function summary #2
Torus:Server supports a number of SQL based functions:
• Markush enumeration
– Typical enumeration speeds > 100k structures per second
• Query format conversion
– MDL MOL/SKC to Daylight SMARTS
• Miscellaneous utility functions
Markush Structures – current work
• Currently support substituent variation (s-variation) – useful
for combinatorial chemistry
• Current and future work extends the Markush representation
to cover further generalisations leading to a full Markush
structure search for use in Patent systems:
– Substituent variation (s-variation)
R1
• R1 can have a number of substitutions
(CH 2) m H
– Frequency variation (f-variation)
• (CH2) repeats m times
– Position variation (p-variation)
• R2 has a number of attachment points
– Homology variation (h-variation)
• R3 is Alkyl, Methyl
R3
R2
Torus:Patent prototype demo
• Live demonstration based up Thomson patent data, all
data held on laptop
• Schema contains patent PDF, patent ID and patent
Markush structures
• Database covers 8,709,704,764,382,349 specific
compounds if enumerated
• Oracle/Torus:Patent server running on laptop
Query creation
• Queries against Torus:Patent database are easily drawn in
ISIS/Draw (supporting all ISIS Query features) and transferred back
into Torus:View ready for query against the patent database in
Torus:Patent
Result viewing
• Results of the query show patent structure along with the original
patent in PDF format in a list
Result viewing
• Patent chemistry structures can also be displayed. A viewer is
available which allows easy viewing of all Markush core/R-Group
Markush core(s)
members
Tree view of
Markush structure
R-Group members
Result viewing
• The original PDF of each result patent can be viewed directly from
Torus:View
Contact Us

www.digitalchemistry.co.uk

[email protected]

+44 (0)113 2181850

The Iron Shed, Harewood House Estate,
Leeds, LS17 9LF, United Kingdom