Initial Strategy and Design
Create molecule type
-
opaque type containing two fixed length strings and one variable length string
Contains number of bits set in the fingerprint, fingerprint(1024 or 2048 bits), SMILES
Allows for 30 kbyte size SMILES (or reactions SMILES)
applications only "see" the SMILES, in <-> out is SMILES
Create Functions which build merlin capabilites into the IUS/UD without a merlin server
Exact lookup
- function overload the SQL "Equals" for tmol datatype:
select molecule from mytable
where molecule
=
"C1CCCCC1";
RETURNS--->
C1CCCCC1
Similarity Searching
with both structures and scores returned:
select molecule, tani from mytable
where
day_tani7
(molecule,'c1ccccc1Br',tani # real);
RETURNS--->
c1ccccc1Br 1.0
c1ccccc1 0.8
with structures above a user defined threshold returned:
select molecule from mytable
where
day_tani
(molecule,'c1ccccc1Br') > 0.5;
reaction similarity searching:
select molecule from mytable
where
day_tani(
molecule,'NNc1cccnn1>>c1ccnnc1') > 0.7;
Substructure Searching
SMILES query
select molecule from mytable
where
day_findassub
('c1ccccc1',molecule);
Reaction SMILES query
select molecule from mytable
where
day_findassub
('c1ccccc1>>', molecule);
SMARTS query
select molecule from mytable
where
day_findassub
('c1ccccc1[Cl,Br]');
Superstructure Searching
Same as Substructure except use the function
day_findassuper
Example
serial interface- Reaction substructure search
Package it up into SQL scripts and shared library "day.bld"
SQL scripts
cansmi toolkit program with Informix API
DBDK Blade Manager- Drag and Drop GUI
Schematic
EuroMug'98, 6-7 October 1998 Cambridge, England
Daylight Chemical Information Systems, Inc.
info@daylight.com