Data processing: database and file management or data structures – Database design – Data structure types
Reexamination Certificate
1999-04-28
2004-04-13
Allen, Marianne P. (Department: 1631)
Data processing: database and file management or data structures
Database design
Data structure types
C702S019000, C702S027000
Reexamination Certificate
active
06721754
ABSTRACT:
BACKGROUND OF THE INVENTION
1. Field of the Invention
The present invention relates generally to information databases, specifically to database similarity joins and more specifically to a system and method for information organization whereby characteristics regarding entities are inferred from the characteristics of similar entities. This is referred to herein as a “fuzzy similarity join” and is exemplified using a chemical similarity join.
2. Background Information
A. Development of Drug Candidates
Chemists, biologists and other users regularly create and test series of chemical compounds in investigating and verifying a hypothesis. In this process, the users often seek to obtain chemical compounds exhibiting certain characteristics, or behaving according to certain metrics, and may seek to synthesize compounds having similar characteristics or behavior patterns.
The process of searching for a chemical compound of some commercial value usually starts with broad-based selection and testing. An example of this is the high-throughput screening typically used in the initial phase of pharmaceutical agent discovery. Pharmaceutical discovery is used as an example, but the same type of process is used for agricultural chemical discovery and material science research, as well as in other related fields.
In High-throughput Screening (“HTS”), the number of compounds examined and tested for a desirable biological response can often range from 50,000 to 500,000, or more. The goal is to find some smaller set of compounds within the larger set that are active in a biological screen, and to treat these compounds as “leads” that can be further developed into an eventual drug candidate. The initial library of compounds tested represents many different types of chemicals.
The chemicals in the initial library can come from several sources, including those developed in-house by conventional synthesis, commercial acquisition, combinatorial chemistry, and natural product extraction. These compounds are typically placed in micro-titer plates. Typical formats for the plates include 96 and 384 well plates, but there is a trend to higher-density plates such as 1536 and 3456 well plates. These plates are typically manipulated by robots to perform the biological screening.
The screens themselves are usually based on a biological receptor. The receptor is either isolated so that binding to the receptor can be measured somewhat directly, or a cell line is engineered to give a detectable response when the receptor is modulated by the potential drug lead.
Although most initial libraries comprise thousands of chemical compounds, even the most extensive library represents a mere sub-set of the trillions (or more) of potential chemical structures that might have “drug-like” characteristics. It is estimated that the total of all compounds available from commercial vendors is currently limited to about 1 million compounds.
The list of compounds to be screened can be selected randomly from those available, or is often chosen with some intuitive “bias” of the chemists or biologists involved in a particular project. This bias can often be advantageous to the project in that chemists often have unique insights into the types of chemicals that may lead to viable drug candidates. However, and as with any bias, an intuitive approach can at times result in potential novel chemicals being overlooked.
In the last few years, the trend in the art has been to select compounds based on the diversity of the compounds within the final selection set. This process is intended to insure that many broad classes of compounds are tested. Both the measure of diversity (diversity metric) and the diversity selection method have been much discussed, but these always are dependent on a measure of similarity between two compounds. The general tendency is to choose compounds that are as different from each other as possible, but this can often lead to selection of the most chemically “unique” compounds in the set; accordingly, this approach can lead to overlooking or missing potentially active lead compounds.
In conducting these studies, researchers rarely desire selection methods that find large clusters of structurally similar compounds within the library (e.g. 5000 benzodiazepam derivatives would not be desirable). Singletons, i.e., compounds that have no similar structure in the dataset, are also generally considered undesirable because these do not allow for the opportunity to develop any structure-activity correlation information. Rather, selection methods that lead to sets of 10-15 similar structures are considered preferable. Such small sets of similar compounds allow for some analysis of the effect of small structure variations on the activity of the compounds (referred to as Structure-Activity Relationships, or, SAR studies). In addition, the small clusters help validate the screening—if 5 of 10 compounds in a small cluster evidence biological activity, because the cluster is comprised of chemically related structures, the activity is more likely to be reproducible and “optimizable.”
The initial biological screening produces compounds that are generally referred to as “chemical hits” or simply “hits”—hits are compounds that have been screened in an assay and evidence biological activity above a desired threshold. These hits rarely include the final drug candidate that will be further analyzed in animal toxicology studies and, ultimately, in human clinical trials. Indeed, these hits generally represent leads that are optimized by producing small changes in their chemical structures; these changes are generally intended to improve or enhance the biological activity of the leads until a commercial candidate is identified via additional screening. These follow-on compounds can be referred to as analogues of the initial hits. This process of optimization of the hits is generally referred to as “lead follow-up.”
Lead follow-up has generally been accomplished by medicinal chemists, who make small sets of analogues of some of the lead compounds. As with the initial screen that led to the initial hits, the analogues are then also tested for biological efficacy. The structure modifications that resulted in reduced activity are usually discarded in favor of those that increased the activity, and new modifications to the analogue compounds are often also made and tested. The medicinal chemist follows the leads until a compound (or a small set of compounds) is identified that has appropriate efficacy for a drug candidate.
In the last several years, the medicinal chemist has often been aided by computer-based design technologies such as Quantitative Structure Activity Relationships (QSAR). These programs use efficacy data for previously tested compounds to predict the efficacy of compounds yet to be tested. The goal of QSAR program is to give accurate predictions of the activities prior to testing the compounds. QSAR programs have generally been successful, not in predicting the activity of the eventual drug candidate, but in allowing more efficient selection of each round of analogue synthesis. While the compounds predicted to be active by QSAR methods do not always have the activity predicted, generally these compounds have an increased chance of being active compared to the general population.
Pharmaceutical development is generally very competitive. Therefore, and almost without deviation, once a drug candidate is selected, extensive patent searches are conducted in order to insure that the candidate if or the use of the candidate is not restricted by another's patent position. Animal toxicology studies generally follow the patent search. If the animal toxicology results are acceptable, human clinical trials of the drug candidate are pursued.
The process of screening, analoging and identification of potential drug candidates can be very time consuming and expensive. Patent searching, particularly in the area of chemical compounds, can also be very time consuming and expensive. Animal toxicology studies involving the potential drug ca
Hurst John R.
Hutton Scott G.
Allen Marianne P.
Arena Pharmaceuticals Inc.
O'Connor Cozen
Straher Michael P.
Wieczorek Ann A.
LandOfFree
System and method for database similarity join does not yet have a rating. At this time, there are no reviews or comments for this patent.
If you have personal experience with System and method for database similarity join, we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and System and method for database similarity join will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-3219480