| Molecular interaction predictors -> Monitor Keywords |
|
|
 
Molecular interaction predictorsMolecular interaction predictors description/claimsThe Patent Description & Claims data below is from USPTO Patent Application 20070192037, Molecular interaction predictors. Brief Patent Description - Full Patent Description - Patent Application Claims
CROSS REFERENCE TO RELATED ALLPLICATIONS [0001]This application is a divisional of U.S. patent application Ser. No. 11/356,196, filed on Feb. 16, 2006, and entitled MOLECULAR INTERACTION PREDICTORS. This application is related to U.S. patent application Ser. No. ______ , (MS315986.02/MSFTP1289USA) filed on Oct. 3, 2006, and entitled MOLECULAR INTERACTION PREDICTORS. The entireties of the aforementioned applications are incorporated herein by reference. BACKGROUND [0002]Living organisms possess various mechanisms for preventing disease states. For instance, the vertebrate immune system provides both humoral-mediated and cellular-mediated immunological defenses. As part of the cellular arm, cytotoxic CD8+ T cells kill infected cells if they recognize short peptides (amino acid subsequences) from a pathogenic protein, which are presented within the Major Histocompatibility Complex class 1 (MHC-1) molecules on a cell's surface. Most human cells create such short peptides by a process that trims proteins down to a length of 8-11 amino acids suitable for binding to MHC-I molecules, or around 20 amino acids suitable for binding to MHC-II. The MHC molecules bind to some of the processed peptides (referred to as epitopes) and present them on the surface of the cell where the cells of the immune system can encounter and recognize the epitopes. The particular epitopes that can be presented by a cell depend on the type of MHC molecules expressed by the organism. [0003]The human MHC molecules are also often referred to as the Human Lymphocyte Antigen (HLA) molecules. MHC-I (HLA-1) molecules are encoded in three regions of the human genome, labeled A, B, and C. Since each individual inherits genes from two parents, each individual expresses from three to six different MHC molecules. The regions of the genome that code for MHC molecules are among the most variable in the human genome. The diversity is concentrated in those nucleotide sequences coding for the groove region of the MHC molecule where an epitope binds to the MHC molecule. [0004]Since different MHC molecules typically bind to different peptides, it is very important clinically to classify MHC types. For example, organ transplant recipients may reject organs received from donors with different MHC types because the cells in these transplanted organs will present MHC-peptide complexes that are new to the immune system of the recipient. Modern MHC typing is performed by sequencing, and the sequence data for all known MHC variants is publicly available. [0005]The interaction between an MHC molecule and a peptide (or any two molecules) can be characterized by a binding free energy. The lower the binding free energy, the greater the affinity between the two proteins. The binding free energy is the difference between the free energy of the bound and unbound states. The binding energy for an MHC-peptide complex can be directly measured by competition experiments with a standard peptide. It is expressed as the ratio between the half-maximal inhibitory concentration (IC50) of the standard peptide to that of the test peptide. In the context of MHC-peptide binding, IC50 is the concentration of the test peptide required to inhibit binding of the standard peptide to MHC by 50%. The result of such experiments is a set of relative binding energies (negative logarithms of the relative concentrations), for different MHC-peptide combinations. [0006]Despite significant progress over the last few years, predicting 3-D protein structure and binding remain difficult to solve problems. Research in this area has focused on complex physics-based models using a large number of particles to describe not only the amino acids in the proteins, but also the solvent that surrounds them. One example of a structural model that can be used to predict peptide-MHC affinity is the threading model. The threading model is based on the premise that proteins fold in a finite number of ways and that the change in the short peptide that binds to MHC does not dramatically influence the 3-D binding configuration. Therefore, instead of screening all theoretically possible ways a particular sequence can fold and bind to another peptide to properly choose the sequence's 3-D structure, the protein binding configurations that are already known are used to compute the binding energy (or affinity). [0007]Due to the importance of MHC complexes, many structures of MHC-peptide binding configurations have been obtained by crystallographers. Since x-ray crystallography reveals that MHC-peptide complexes exhibit a finite number of conformations, the threading approach can be applied to the problem of predicting MHC-peptide binding. The threading approach assumes that energy is additive, but it introduces a simplification that allows estimation of the binding energy of a peptide with an MHC molecule whose 3-D configuration of binding with some other peptide is known. In particular, the assumption is that the binding energy is dominated by the potentials of pairwise amino acid interactions that occur when the amino acids are in close proximity (e.g., distance smaller than 4.5 .ANG.). Another assumption underlying the threading approach is that the proximity pattern of the peptide in the groove (i.e., MHC binding site) does not change dramatically with the peptide's amino acid content. As the pairwise potentials are assumed to depend only on the amino acids themselves and not on their context in the molecule, the energy becomes a sum of pairwise potentials taken from a symmetric 20.times.20 matrix of pairwise potentials between amino acids. These parameters are computed based on the amino acid binding physics and there are several published sets derived in different ways. [0008]The MHC-peptide threading procedure utilizes solved MHC-peptide complexes as the threading template, a definition of interacting residues and a pairwise contact potential table. To predict MHC-peptide binding, the query sequence is "threaded" through the various known MHC structures to find the best fit. These structural data files are available, for instance, from the Research Collaboratory for Structural Bioinformatics (RCSB) protein data bank (www.rcsb.org/pdb). The algorithm for the threading model proceeds as follows. Given a known structure of an MHC-peptide complex, the contacting MHC residues for each peptide position are determined. The amino acid-amino acid pairwise potentials are used to score the interaction of a peptide amino acid at a certain position with all its contacting residues. Assuming position independence, the peptide's score is the sum of the amino acid scores. [0009]An example of an MHC-peptide complex is given in FIG. 1, which shows the 3-D structure of MHC A0201 bound to a peptide. The peptide amino acid centroids are marked in 3-D space by triangles and the centroids of the MHC amino acids are marked by circles. The MHC amino acids that are in proximity (<4 .ANG.) of the peptide are marked by filled circles. SUMMARY [0010]This Summary is provided to introduce a selection of concepts in a simplified form that are further described below in the Detailed Description. This Summary is not intended to identify key features or essential features of the claimed subject matter, nor is it intended to be used to limit the scope of the claimed subject matter. [0011]The subject matter described herein utilizes machine learning techniques to predict molecular interactions. By way of example, a threading model can be expressed as one or more parametric functions having learnable parameters. The parameters can be estimated from available data and the trained adaptive threading model can be used to predict molecular interactions. The available data can be of any type suitable for the particular molecules under study. For instance, if the adaptive threading model is used to predict protein-protein (e.g., MHC-peptide, receptor-ligand, antibody-antigen, etc.) binding energies, the parameters can be estimated from the protein sequences, 3-D protein-protein complex structural data and known binding energies (continuous or threshold) for similar protein-protein complexes. Knowledge of epitopes includes knowledge of threshold or binary energy data. For peptides that are epitopes, this implies low binding energy. If the IC50 or some other continuous measure of binding affinity is not known, other sources of binding data can be used. For instance, for some peptides, the information in the published literature can be used to determine whether they are or are not epitopes even though their exact binding energy is not known. In these cases, binary (low or high) information about binding energy can be used. [0012]The adaptive model can infer unknown data using machine learning techniques. The learnable parameters can be, for instance, contact potentials, weights and distance function parameters. Any suitable machine learning technique can be used to estimate the parameters and infer unknown variables and parameters so as to maximize the fit of the known data to the model (e.g., iterative optimization, iterative least squares, expectation maximization (EM), generalized expectation maximization (GEM), variational expectation maximization (VEM), gradient descent, conjugate gradient descent, etc.). [0013]The subject matter can be used to not only identify molecules with very low binding energies (i.e., good binders such as epitopes), but also to rank the molecules having intermediate levels of binding. The model significantly outperforms the standard threading approach in binding energy prediction. The subject matter also can be used to identify the effects of host immune pressure on pathogen evolution (e.g., HIV sequence evolution within a host and on a population level). [0014]The following description and the annexed drawings set forth in detail certain illustrative aspects of the subject matter. These aspects are indicative, however, of but a few of the various ways in which the subject matter can be employed and the claimed subject matter is intended to include all such aspects and their equivalents. For ease of description, MHC-peptide binding energies have been selected to illustrate how the subject matter can be employed. However, the subject matter facilitates making predictions about any molecular binding configuration, especially protein-ligand binding or other situations in which a family of similar molecules have been documented to bind to a variety of molecules of interest and is not limited only to predicting MHC-peptide binding energies. A ligand can be any molecule (especially a small molecule, such as a peptide) that binds to another. BRIEF DESCRIPTION OF THE DRAWINGS [0015]FIG. 1 schematically illustrates the 3-D structure of MHC A0201 bound to a peptide. [0016]FIG. 2 is a block diagram of one example of a system that facilitates making a prediction relating to a molecular interaction. [0017]FIG. 3 is a block diagram of another example of a system that facilitates making a prediction relating to a molecular interaction. [0018]FIG. 4 is a block diagram of another example of a system that facilitates making a prediction relating to a molecular interaction. [0019]FIG. 5 is a block diagram of another example of a system that facilitates making a prediction relating to a molecular interaction. [0020]FIG. 6 is a block diagram of yet another example of a system that facilitates making a prediction relating to a molecular interaction. Continue reading about Molecular interaction predictors... Full patent description for Molecular interaction predictors Brief Patent Description - Full Patent Description - Patent Application Claims Click on the above for other options relating to this Molecular interaction predictors patent application. Patent Applications in related categories: 20090287419 - Agglutination judgment method - The present invention provides an agglutination judgment method includes a first step of collecting data of a transmitted light intensity from each of measuring points included in a predetermined region of the reaction pattern, a second step of calculating a parameter representing the dispersion of a transmitted light intensity in ... 20090287417 - Method for detecting a biochemical interaction - A method for detecting a biochemical interaction between at least two interaction partners, comprising the steps of bringing into contact the at least two interaction partners, taking a temporal sequence of measurements, each of them producing a measurement value describing the state of the interaction at a given point in ... 20090287418 - Population of background suppression lists from limited data in agent detection systems - Methods and systems are disclosed for detection of agents such as pathogens or toxic substances and, in particular, to methods and systems for determining the most important background constituents to suppress in a sample, e.g., in a bulk aerosol sample, in order to reduce the probability of false alarms and ... ###  How KEYWORD MONITOR works... a FREE service from FreshPatents1. Sign up (takes 30 seconds). 2. Fill in the keywords to be monitored. 3. Each week you receive an email with patent applications related to your keywords. Start now! - Receive info on patent apps like Molecular interaction predictors or other areas of interest. ### Previous Patent Application: Molecular interaction predictors Next Patent Application: System for providing vehicular hospitality information Industry Class: Data processing: measuring, calibrating, or testing ### FreshPatents.com Support Thank you for viewing the Molecular interaction predictors patent info. IP-related news and info Results in 0.27429 seconds Other interesting Feshpatents.com categories: Daimler Chrysler , DirecTV , Exxonmobil Chemical Company , Goodyear , Intel , Kyocera Wireless , 174 |
 * Protect your Inventions * US Patent Office filing 
PATENT INFO |
|
