This document contains the following major sections:
Introduction
ChemMart
is a ChemDraw Interface to Searle Computational Chemistry Tools
that allows you to sketch in molecules or import them from your
existing ChemDraw files. You can then select from several
computational tools maintained by the
Drug Design Group at Searle
to find similar structures in the available databases or to compute
chemical properties for the structures you entered.
Push one of the buttons to the right of the ChemDraw sketching area
to perform a search or compute molecular properties.
The available tools are:
Searching Tools
Search strategies
The available databases are MDL's ACD, the Searle SC file,
and the Monsanto CP and CAP databases. Using the options
buttons allows you to select one or more of these databases
for searching. You may also set certain parameters, such as
the similarity index for neighbor searhcing by using the
options button. The neighbor, substructure and SMARTS searches
also allow you to refine(filter) your search to allow only compounds
in certain molecular weight ranges, from certain suppliers (ACD
only), and to allow or disallow compounds containing
particular funtional groups. The Search/Filters Options
is used to define these filters.
Find structure
The Find structure search is the simplest. It looks into
each of the databases you have selected and finds (or not) the
exact structure(s) you have entered.
Substructure
The Substructure search locates all structures in the selected
databases which contain the structure you have entered. This may
return the exact structure as well, if it is contained in the
database(s). It also reports the similarity of the returned
structures to the input structure. This similarity index
is the same as that used in the Tanimoto method
described under Neighbor searching.
Neighbors
The Neighbors search locates structures in the database(s)
by computing a
similarity measure of the structure(s) you have
entered with each structure in the database(s).
The similarity measure is a number between 0 and 1. A value
of 1 generally means the structure is very similar and
likely identical to the input structure. You may select
the similarity threshold to indicate how similar the
returned structures should be to your input structure.
The default similarity threshold is 0.7. In other words,
structures having a similarity measure at least 0.7 are found.
Use the following guidelines if you wish to change this threshold.
There are two similarity methods which you can choose from when doing a Neighbor search: Tversky and Tanimoto, named after people who described the algorithms used to compute the similarity index. Both methods use fingerprints which contain information about the presence or absence of sub-structural moieties.
Tversky method
The
Tversky method is the default method for neighbor searching.
It uses two additional
parameters, alpha and beta. Roughly speaking, these are
the relative weight to be assigned to the presence of
similar moieties(alpha) and the absence of them(beta).
ChemMart uses alpha=1.0 and beta=0.2 by default, but
you may change these. The default search therefore favors
the selection of structures which contain similar moieties
even if they also contain (somewhat) disimilar ones.
Note: using the default alpha and beta, a returned similarity index
of 1.0 generally indicates that the found structure contains
the exact substructure of the structure you have entered,
although it may also contain (many) other disimilar moieties.
To entertain yourself, you may set alpha=0.0 and beta=1.0
and discover structures which are the most unlike your
input structure(s).
Tanimoto method
When you use the
Tanimoto method for neighbor
search the structures located will contain the greatest
percentage of similar moieties AND the fewest percentage
of disimilar moieties. It is roughly the same as the
Tversky method with alpha=beta=1.0.
SMARTS
SMARTS is a way of performing substructure searches that
are much more general than the simple substructure search explained above.
The simple substructure search only allows you to find structures
which exactly contain the specific structure which you have
enterted. Using SMARTS, you could also find those
structures which contain other related structures, for example
with wildcard atoms and bonds, or specific set of atoms at some
particular ring position. Using the ChemDraw plugin
and a few hints,
you can specify a wide variety of SMARTS searches.
In order to do a completely general SMARTS search, you must understand the SMARTS language. When you press the SMARTS button in ChemMart, you a presented with a page allowing you to modify the SMILES string representing the structure you have entered. You should change it into a valid SMARTS. Of course, you can completely erase the SMILES and enter any valid SMARTS. If you simply proceed without changing the SMILES, you will be doing a simple substructure search for the structure you have entered, but it will take about twice as long compared to the ChemMart standard substructure search.
The ChemDraw plugin
ChemMart uses a special
web version of ChemDrawPro v4.5
which functions as a plug-in to your web browser.
In order to use this, you must have installed this plug-in.
The plug-in web version of ChemDraw should be familiar to users of regular ChemDraw, but a tutorial is avalable for new users. The drawing area to the left of the action buttons is where you sketch in your molecules. The vertical toolbar is part of the ChemDraw web application and should be familiar to ChemDraw users. The traditional horizontal menu at the top of ChemDraw is now implemented using a pop-up menu. Use the right mouse button in the drawing area to pop up this menu.
You may either sketch in your molecules, or import a ChemDraw file you have previously created. If the file contains multiple molecules, results will be returned for all molecules when you choose one of the action buttons on the right.