(Molecular Descriptors): Hidden Layer

(Molecular Descriptors): Hidden Layer

US 200801 04001A1 (19) United States (12) Patent Application Publication (10) Pub. No.: US 2008/0104001 A1 Kipp (43) Pub. Date: May 1, 2008 (54) ALGORITHM FORESTIMATION OF Related U.S. Application Data BINDING EQULIBRLA IN INCLUSION COMPLEXATION, HOST COMPOUNDS (60) Provisional application No. 60/863,296, filed on Oct. IDENTIFIED THEREBY AND 27, 2006. Provisional application No. 60/896,021, COMPOSITIONS OF HOST COMPOUND AND filed on May 15, 2007. PHARMACEUTICAL Publication Classification (76)76 Inventor: James E. Kipp, Wauconda, IL (US) (51) G06N,Int. Cl. 3/08 (2006.01) (52) U.S. Cl. ................................................................ 706/25 Correspondence Address: (57) ABSTRACT BAXTER INTERNATIONAL INC. - 0 One Baxter Parkway, DF2-2W The present invention discloses a neural network and associ Deerfield, IL 60015-4633 (US) ated algorithms for improving the identification of chemi cally useful compounds without having to test each investi gated compound individually. The method utilizes a neural (21) Appl. No.: 11/923,775 network and associated algorithms for estimating the ability to dissolve poorly water soluble molecules by formation of (22) Filed: Oct. 25, 2007 water-soluble inclusion (guest-host) complexes. input layer (molecular descriptors): Hidden layer: Output layer (estimate): Patent Application Publication May 1, 2008 Sheet 1 of 5 US 2008/O104001 A1 FIG. 1 OH HO 9 O O OH HO HO O HO O OH HO HO O OH OH O OH OH HO O OHO OH HO HO O O HO FIG. 2 Input layer (molecular descriptors): Hidden layer: Output layer (estimate): Patent Application Publication May 1, 2008 Sheet 2 of 5 US 2008/01 04001 A1 FIG. 3 FIG. 4 0.75 0.7 0.65 0.6 0,55 O.5 0.45 0.4 SR 2 C SE g s cv. s S s s i Descriptor Patent Application Publication May 1, 2008 Sheet 3 of 5 US 2008/01 04001 A1 FIG. 5 0.705 0. 7 0.695 0. 6 9 0.685 0.68 0.675 0.67 0.665 5 7 9 11 13 15 17 Number of hidden neurons FIG. 6 R2 - 0.791 o(reg) = 0.347 -1.0 00 10 2.0 3.0 4.0 5.0 6.0 Measured K(1:1) Patent Application Publication May 1, 2008 Sheet 4 of 5 US 2008/O104001 A1 FIG. 7 R2 = 0.703 o(reg) = 0.441 1.0 00 1.O 2.0 3.0 4-O 5.O 6.0 Measured K(1:1) FIG. 8 6.O R2 = 0.809 5.0 o(reg) = 0.345 3 O o -1.0 0.0 1.0 2.0 3.0 4.0 5.0 6.0 Measured K(1:1) Patent Application Publication May 1, 2008 Sheet 5 of 5 US 2008/O104001 A1 FIG. 9 R2 = O.800 o(reg) = 0.499 0.0 1.O 2.0 3.0 4.0 5.0 Measured K(1:1) FIG 10 R2 c 0,623 O 5 10 15 Ranking (maximum decrease) US 2008/01 04001 A1 May 1, 2008 ALGORTHM FORESTIMATION OF BINDING 0005 High-throughput systems and computational meth EQULIBRLA IN INCLUSION COMPLEXATION, ods have been proposed in the area of drug discovery (Braun HOST COMPOUNDS IDENTIFIED THEREBY AND heim, U.S. Pat. No. 6,587,845, entitled “Method and Appa COMPOSITIONS OF HOST COMPOUND AND ratus for Identification and Optimization of Bioactive PHARMACEUTICAL Compounds. Using a Neural Network”). This patent discloses the removal of a single value from a neural network type of CROSS-REFERENCE TO RELATED training set, used as an “adjuster. This value is left out during APPLICATIONS training and used to check if the neural network generalizes and is not overtrained. The neural network is trained until 0001. This application claims the benefit of U.S. Provi convergence, and the error between the actual and predicted sional patent applications, Ser. No. 60/863.296 filed Oct. 27. output for the adjuster value is calculated. If the neural net 2006 and Ser. No. 60/896,021 filed May 15, 2007. The entire work predicts the adjuster value within 5%, that neural net text of each of the aforementioned applications is incorpo work's construction is saved. If the prediction is more than rated herein by reference. 5% off, a new network is chosen. However, the process for BACKGROUND OF THE INVENTION choosing a new network is not defined. In the 845 patent, this procedure is repeated until a construction is found that allows 0002. A computational method is disclosed herein for the neural network to predicta target to within 5%. The 845 development of chemically useful agents in solution. A neural patent further states that the “most common neural network network and associated algorithms are disclosed and used for construction is chosen as the final construction and that the estimating the ability to dissolve poorly water soluble mol final construction for this system is five hidden layer neurons, ecules by formation of water-soluble inclusion (guest-host) ten thousand iterations, learning rate equals 0.1 and the complexes. This method is an improvement over current momentum term equals 0.9". methods of new product development, which often rely upon experimental trial and error, a time-consuming and costly 0006 The present disclosure is an improvement on this process. The computational method embodied in this disclo type of art, which does not provide a method for developing Sure predicts specific material properties, the knowledge of an entire network structure; that is: number of input param which facilitates the design of useful aqueous solutions. The eters, number of hidden layers, number of neurons per layer, present neural network is “trained on known binding con and so forth. There is thus a need, provided by the present stants for complexation of guest molecules with a host mol disclosure, to develop a more complete network structure. ecule such as cyclodextrin, and then used to predict the bind There accordingly is a need for approaches that are oriented ing affinity of unknown compounds. Special applications toward formulation development, and more particularly that include developing host compounds, such as specific cyclo are designed to estimate the ability to Successfully formulate dextrin compounds, for preparing pharmaceutical composi compositions of water-insoluble compounds by use of inclu tions with advantageous coordination between properties of sion complexation and specific solubilizing agents. the host compound and the pharmaceutical. 0007 Inclusion complexation is a process of rendering FIELD OF THE INVENTION insoluble compounds more water soluble by enclosing the less water-soluble compound (guest molecule) within a cav 0003. This invention pertains to the field of using compu ity of the soluble “host compound. Examples of such host tational methods in predictive chemistry. More particularly, compounds include cyclodextrins and their derivatives. the invention utilizes a neural network with associated algo Cyclodextrins comprised of six to eight glucopyranoside rithms, and the known properties of the molecules investi units assume a toroid structure, or truncated cone with the gated, to optimize the prediction of physical properties for ends consisting of a large diameter rim and Small diameter molecules of interest. rim. Dissolved in water, the hydroxy groups on these rims are 0004 Traditional development of chemically useful solu exposed to the aqueous environment. This configuration tions, such as those in the pharmaceutical art, have involved causes the interior of the cyclodextrin to be considerably less the arduous task of preparing test formulations in the labora hydrophilic than the aqueous environment and thus able to tory and conducting stepwise experiments to elucidate perti interact with other hydrophobic molecules. The exterior is nent chemical properties. These properties may include, but sufficiently hydrophilic to render cyclodextrins and their are not limited to, the following: solubility, water-oil parti complexes more water Soluble than the hydrophobic guests. tioning, water-n-octanol partitioning, chemical stability, and 0008 The formation of the inclusion compounds greatly physical stability, which are known to affect the ability to modifies the physical and chemical properties of the guest formulate a product. In the pharmaceutical industry, this slow, molecule. Most importantly, the water solubility is enhanced. costly throughput is compounded by a lengthy drug discovery For this reason, cyclodextrins have attracted interest in many process, in which historically over 10,000 compounds must fields, and have led to the production of many chemically be individually tested and evaluated for every one that actu useful products. For example, a commercially available ally reaches the market (SCRIP World Pharmaceutical News, deodorizing solution of Proctor & Gamble is composed Jan. 9, 1996, PJB Publications). Many times this failure can largely of a cyclodextrin in an aqueous medium. The "dryer be attributed to water insolubility, which limits administra sheets” that are used to release pleasant scents when laundry tion by a therapeutically effective route. This stark realization is heated are fabric or paper that is impregnated with dry, Solid has driven many research organizations to shift their focus cyclodextrin microparticles that have been exposed to fra from traditional drug discovery to development of high throughput systems (HTP), or computational methods that grances. leverage computer technology in the drug discovery and 0009 Cyclodextrins can also be used in environmental development process. decontamination because they can effectively immobilize US 2008/01 04001 A1 May 1, 2008 toxic compounds inside their rings. For example, trichloroet weighted Sum from neurons in the previous layer applies a hane, trichlorfon (an organophosphorus insecticide), and transfer or activation function to this sum. In practical terms, heavy metals, among many other compounds can form inclu neural networks are non-linear statistical modeling tools that sion complexes with cyclodextrins. Cyclodextrins are can be used to model complex relationships between inputs employed in the production of cholesterol free food products, and outputs, or to find data patterns.

View Full Text

Details

  • File Type
    pdf
  • Upload Time
    -
  • Content Languages
    English
  • Upload User
    Anonymous/Not logged-in
  • File Pages
    18 Page
  • File Size
    -

Download

Channel Download Status
Express Download Enable

Copyright

We respect the copyrights and intellectual property rights of all users. All uploaded documents are either original works of the uploader or authorized works of the rightful owners.

  • Not to be reproduced or distributed without explicit permission.
  • Not used for commercial purposes outside of approved use cases.
  • Not used to infringe on the rights of the original creators.
  • If you believe any content infringes your copyright, please contact us immediately.

Support

For help with questions, suggestions, or problems, please contact us