Molecular hologram QSAR

Details Molecular hologram QSAR Molecular hologram QSAR

1998-02-10

2001-03-27

Assouad, Patrick (Department: 2857)

Data processing: measuring, calibrating, or testing

Measurement system in a specific environment

Chemical analysis

C702S031000, C700S268000, C700S293000

Reexamination Certificate

active

06208942

ABSTRACT:

BACKGROUND OF THE INVENTION
1. Field of the Invention
This invention relates to the field of chemical discovery and to understanding the relationship between the structure of a molecule and its chemical function (a structure/function relationship) and especially as structure/function relationships relate to biological chemical discovery in the search for new medicinal drugs. In particular, a method has been discovered that uses an iterative process of determining, using the partial least squares method of multivariate analysis, which definition of a specialized 2D fragment molecular metric best characterizes the structure-activity relationship among a series of molecules having similar activities. Once identified, this definition can be used to visualize, in a computer graphics environment, the relative contributions of each portion of a molecule to its chemical activity.
2. Description of Related Art
1. Structure-Activity Relationships
In the never-ending search for new and more effective drugs with which to treat disease, one approach to discovery has been the mass screening of naturally occurring chemical compounds. More recently, huge schemes of combinatorial chemical synthesis have produced mass numbers of additional chemical compounds available for screening. However, once an active chemical is identified, a search must still be conducted to find the molecular relative of the identified molecule which has the greatest activity in the desired biological system. One of the principal techniques employed by medicinal chemists has been to examine the chemical structures of a series of molecules which are related by the fact that they all exhibit some activity in the biological system of interest, and, relying on fundamental chemical and physical principles, making educated guesses as to which part or parts of the molecules is/are most important to the activity. Based on these guesses, new compounds can be synthesized and tested.
Over the years quantitative approaches to relating structure and activity were developed to supplant the intuitive guess of chemists. These approaches generally sought to cast the observed/measured biological value (Ob) in terms of a linear combination of molecular descriptors A, B, C, etc. [Ob=A+B+C . . . (n)] Thus, for each of the molecules which are related by the fact that they all exhibit some activity in the biological system of interest, a row is entered in a data table (matrix) for that molecule as shown in FIG.
1
. Unless (which was rarely, if ever, the case) the number of molecules equaled the number of descriptor values, an inherently underdefined system of equations was presented, and no explicit solution could be found. Various molecular descriptors were developed to characterize the molecules having similar activities and a relationship was sought by applying various statistical methods of analysis (such as multiple linear regression) to the underdetermined data table.
These systems of “quantitative structure activity relationships” (acronymed QSAR) enjoyed modest success in drug design but generally failed in their attempt to quantitatively take the three dimensional shape of molecules into account, a necessary requirement for biological systems for which the three-dimensional stereo conformation of biomolecules and their substrates has been shown to be of preeminent importance. Ultimately in 1988, a sophisticated method (CoMFA
1
) of comparing the three-dimensional shapes of molecules and relating the shapes to observed biological activity differences to identify the most important common topological features of the molecules was developed. Typically, molecular shape descriptors consisting of thousands of terms were defined for a relatively few molecules. The resulting data table was successfully analyzed using the Partial Least Squares (PLS) statistical technique to extract meaningful structure-activity information. This Comparative Molecular Field Analysis (CoMFA) approach has been remarkedly successful and has enjoyed wide acceptance and usage. However, to use CoMFA, skilled medicinal-computational chemists are required to make difficult and complex decisions regarding molecular conformation and relative alignment and a significant amount of computational time is then required to achieve the full benefits of COMFA.
2. 2D Molecular Fingerprints
Molecular fingerprints are bitmaps representative of a molecule and have been primarily used to efficiently search databases and to analyze chemical similarity
2
. Essentially, a long binary bit string which consists of 0s and 1s is created for each molecule. Each position along the string is assigned to a specific molecular fragment. If that fragment exists in the molecule under consideration, the corresponding bit is set to 1, otherwise it is left as a 0. For the present purposes, two interwoven characteristics of the bit strings are important. First, because of the way in which fragments are defined, the same molecular structure (functional group, atomic arrangement, etc.) may be included in more than one fragment and, thus, contribute to setting more than one bit in the string at 1. As a result of this, more than one unique molecule may specify the same bit string. Put another way, there is an inherent degeneracy in this method so that one can not go backwards to a molecule from a knowledge of its bitmap. Further, despite the fact that fragments must have some relationship to the three dimensional structure of the molecule, that relationship is not explicitly incorporated in the bitmap. Thus, it is generally acknowledged that no information relating to the three dimensional structure is directly encoded is this type of bitmap, and it is, accordingly, referred to as a 2 Dimensional (2D) representation.
Similarity assessments between molecules based on 2D fingerprints are most commonly performed using the Tanimoto coefficient
2
, which compares the number of fingerprint bits in common between pairs of structures. Most recently, a technique has been developed which identifies structural commonalities in sets of compounds
3
. This technique (known as Stigmata) essentially ANDs (in a Boolean sense) the 2D fingerprints (binary bit strings) of the structures in the data set and identifies fingerprint bits held in common across some percentage of the data set.
There are two general methods of 2D fingerprint generation supplied by the companies which develop and promote them. The first, known as the keyed
4
method, and a second known as the hashed
5
method. The keyed method requires a priori substructural definitions for all the fragments that should be searched for during the fingerprint generation process; if a fragment is not specified in the input list, it will not be included in the fingerprint. The hashed method uses a set of rules for generating fragments for fingerprinting. That is; generic rules are applied that define how a chemical structure should be broken down into constituent fragments. The hashed method uses these rules to generate all possible unbranched fragments. Both methods result in a binary bit string (0s or 1s) that encode the presence or absence of particular fragments.
In the past, attempts to use 2D FINGERPRINTS to generate useful QSARs have not been successful no matter what type of correlation scheme was employed. It is believed that this was the case because an insufficient amount of three-dimensional information about the molecules was contained in the essentially two-dimensional fingerprint.
Definitions
2D FINGERPRINTS shall mean a 2D molecular measure in which a bit in a data string is set corresponding to the occurrence of a given length atom fragment in that molecule. Typically, strings of roughly 900 to 2400 bits are used depending on how many different combinations of components are utilized. A particular bit may be set by many different fragments.
MOLECULAR HOLOGRAM shall mean a weighted 2D FINGERPRINT in which all possible fragments are counted with each position in the fingerprint to which each fragment is assigned being weighte

Affiliated with

Also associated with

Rating

Molecular hologram QSAR does not yet have a rating. At this time, there are no reviews or comments for this patent.

If you have personal experience with Molecular hologram QSAR, we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Molecular hologram QSAR will most certainly appreciate the feedback.

Rate now

Comments { 0 }

Profile ID: LFUS-PAI-O-2447153