Graphical representation of
stereochemical information vs. computer representation
Published stereoisomers are
graphically depicted using the solid wedge, broken or dashed wedge, broken line,
solid bar and open wedge conventions. For special cases Fisher and Newman
projection formulas are used. Specialised structure-drawing editors enable
entering of complete structures together with the above mentioned graphic
representations of stereo characteristics or projection formulas and then
translate them into connection tables represented internally as character or
binary strings. These strings comply with pre-defined formats. There are great
many formats currently in use, usually customised by the owner or the producer
of a database. Two formats whose details have been published and which are in
widespread use within the chemical community are ROSDAL [20] and MOLFILE [21],
both now supported by Molecular Design Limited (MDL). The latter, MOLFILE, has
became a quasi standard in computer based chemical information applications.
All formats contain coded
information on the atom and bond characteristic of the structure as well as
specifications of atom topology in a form of their 2D or 3D coordinates. From a
chemical point of view (neglecting here for the time being their role in the
structure�s visualisation) the coordinates are only needed for structures with
chirality centres and/or geometrical stereochemistry. Unique association of a
given asymmetric centre with the correct stereodescriptor, particularly for big,
multi-atom structures, is difficult even for an experienced chemist, not to
mention the programmer designing a reliable algorithm for computer handling of
this task. The difficulties usually derive from the way the structures are drawn
on a paper or on the terminal screen. The graphic depiction of structural
diagrams is often far from the correct parallel projection of the ideal
tetrahedron.
Correct and �clean� graphical representation of
stereochemistry in structures, for most of the known computer algorithms
handling chirality centres geometry [22] leads, in most of the cases, to correct
computer representations, i.e., the algorithm is able to calculate and assign
the corresponding, correct, stereodescriptors. The discrepancies come from the
differences between the strictly �chemical thinking� and �mathematical thinking�
approaches projected into the algorithm. Usually, after detailed analysis, it
can be determined that the algorithm had not simulated one-to-one all the
operations the chemist executed while assigning stereochemistry to a structure.
Known relevant algorithmic methods
of geometrical analysis of stereochemical centres assign some fixed value to the
Z-co-ordinate with its sign (direction) determined by the direction of the coded
stereo bond (+, positive, for solid bars and solid wedges or -, negative, for
broken bars and open wedges). Usually the fixed value is equal to 1 (independent
of the co-ordinates scaling) [22], or some �normalised� constant related to the
spectrum of X- and Y- coordinates. The crucial weakness of this approach is its
ability to also generate results for such cases where manual analysis by a
chemist would reject them as absolutely erroneous or inappropriate for analysis.
The impact of this weakness can be reduced by introducing additional rules and
conventions for coding of stereocentres during data entry of structures [23,24].
|
