Apparatus and method for structure-based prediction of amino acid sequences

Abstract

The present invention provides methods and apparatus for analyzing a protein structure.

1 . A method for analyzing a protein structure, the method being executed in a computer under the control of a program stored in the computer, and comprising the following steps:
 A) receiving, by the computer, a reference structure for a protein, whereby
 the said reference structure forms a representation of a three-dimensional structure of the said protein, and has a global energy E ref , 
 the protein consists of a plurality of residue positions, each carrying a particular reference amino acid type in a specific reference conformation, and 
 the protein residues are classified into a set of modeled residue positions and a set of conformationally fixed residues; 
   B) substituting, by the computer, into the reference structure of step (A) a pattern, whereby
 the said pattern consists of one or more of the modeled residue positions defined in step (A), each carrying a particular amino acid residue type being placed at a specific residue position in the reference structure and adopting a specific conformation, and 
 the one or more amino acid residue types of the said pattern are replacing the corresponding amino acid residue types present in the reference structure; 
   C) optimizing, by the computer, the conformation of the reference structure of step (A) being substituted by the pattern of step (B), whereby
 a suitable protein structure optimization method based on a function allowing to assess the quality of a global protein structure, or any part thereof is used in combination with a suitable conformational search method, and 
 the said structure optimization method is applied to all modeled residue positions defined in step (A) not being located at any of the pattern residue positions defined in step (B), with the proviso that said structure optimization method is not applied to any of said pattern residue positions; 
   D) assessing, by the computer, the energetic compatibility of the pattern defined in step (B) within the context of the reference structure defined in step (A) being structurally optimized in step (C) with respect to the said pattern, by way of comparing the global energy E(p) of the substituted and optimized protein structure with the global energy E ref  of the non-substituted reference structure to obtain an energetic compatibility object energy E ECO (p); and   E) generating and outputting, by the computer, to a display, a user, a readily accessible memory, another computer on a network, for another computer-readable medium, in the form of an energetic compatibility object (ECO), a value reflecting the said energetic compatibility of the pattern together with information forming a full description of the three-dimensional structure comprising the substituted pattern;   
       wherein the reference structure of step (A) is received from a database of protein structures. 
     
     
         2 . A method according to  claim 1 , wherein the reference structure of step (A) further represents the three-dimensional structure of ligands including co-factors, ions or water molecules. 
     
     
         3 . A method according to  claim 1 , wherein the reference structure of step (A) has undergone a further energy optimization step by means of a molecular mechanics gradient method. 
     
     
         4 . A method according to  claim 1 , wherein the reference structure of step (A) assumes a reference amino acid sequence being the wild type sequence of the said protein. 
     
     
         5 . A method according to  claim 1 , wherein the reference structure of step (A) assumes a reference amino acid sequence being the sequence of lowest energy for the said protein. 
     
     
         6 . A method according to  claim 1 , wherein the reference structure of step (A) assumes a reference amino acid sequence being a polyglycine or a polyalanine sequence. 
     
     
         7 . A method according to  claim 1 , wherein the reference structure of step (A) assumes any possible reference amino acid sequence including a random or a dummy sequence. 
     
     
         8 . A method according to  claim 1 , wherein the reference structure of step (A) has been modeled by the structure optimization method used in step (C). 
     
     
         9 . A method according to  claim 1 , wherein the set of modeled residue positions comprises at least one position. 
     
     
         10 . A method according to  claim 1 , wherein the set of modeled residue positions comprises at least three positions. 
     
     
         11 . A method according to  claim 1 , wherein the set of modeled residue positions comprises at least five positions. 
     
     
         12 . A method according to  claim 1 , wherein the pattern of step (B) consists of a single residue position, the said pattern being referred to as a single residue pattern. 
     
     
         13 . A method according to  claim 1 , wherein the pattern of step (B) consists of a pair of residue positions, the said pattern being referred to as a single residue pattern. 
     
     
         14 . A method according to  claim 1 , wherein each of the pattern residue positions defined in step (B) are assigned a naturally occurring amino acid residue type. 
     
     
         15 . A method according to  claim 1 , wherein each of the pattern residue positions defined in step (B) are assigned a naturally or a non-naturally occurring amino acid residue type. 
     
     
         16 . A method according to  claim 1 , wherein each of the pattern residue positions carrying a particular amino acid residue type as defined in step (B) are assigned a fixed conformation retrieved from a protein structure database. 
     
     
         17 . A method according to  claim 1 , wherein each of the pattern residue positions carrying a particular amino acid residue type as defined in step (B) are assigned a fixed conformation which is computer-generated. 
     
     
         18 . A method according to  claim 1 , wherein each of the pattern residue positions carrying a particular amino acid residue type as defined in step (B) are assigned a fixed conformation which is received from a rotamer library, the said pattern being referred to as a rotameric pattern, and wherein the said rotamer library is backbone-dependent, the said library being referred to as a combined main-chain/side-chain rotamer library, but wherein the main-chain information in the library is only used in order to assign a side-chain conformation to each of the pattern residue positions carrying a particular amino acid residue type as defined in step (B) which is structurally compatible with the local main-chain conformation of each corresponding residue position in the reference structure of step (A). 
     
     
         19 . A method according to  claim 1 , wherein the said conformational search method of step (C) includes a Dead End Elimination (DEE) method. 
     
     
         20 . A method according to  claim 1 , wherein the said conformational search method of step (C) includes a Monte Carlo simulation method. 
     
     
         21 . A method according to  claim 1 , wherein the said conformational search method of step (C) includes a genetic algorithm method. 
     
     
         22 . A method according to  claim 1 , wherein the said conformational search method of step (C) includes a mean-field method. 
     
     
         23 . A method according to  claim 1 , wherein the said conformational search method of step (C) includes a molecular mechanics gradient method such as a steepest descent- or a conjugated gradient energy minimization method. 
     
     
         24 . A method according to  claim 23 , wherein the said template residues are not kept fixed but are allowed to vary by at most 2 angstroms in root-mean-square deviation compared to the reference structure of step (A). 
     
     
         25 . A method according to  claim 23 , wherein neither the said pattern residues nor the said template residues are kept fixed but wherein they are allowed to vary by at most 2 angstroms in root-mean-square deviation compared to the substituted reference structure of step (B). 
     
     
         26 . A method according to  claim 1 , wherein the residue positions to which the structure optimization is applied as defined in step (C) are rotameric, in that they are allowed to assume conformations received from a rotamer library in the same way as occurs for rotameric pattern residues. 
     
     
         27 . A method according to  claim 1 , wherein the said energetic compatibility of the pattern is calculated as the difference in global energy between the substituted and optimized protein structure of step (C) and the reference structure of step (A). 
     
     
         28 . A method according to  claim 1 , wherein the said energetic compatibility of the pattern is calculated as the absolute global energy of the substituted and optimized protein structure of step (C), this being equivalent to selecting in step (A) a reference amino acid sequence such as a polyglycine sequence or a dummy sequence. 
     
     
         29 . A method according to  claim 1 , wherein the self-energy of the template is ignored. 
     
     
         30 . A method according to  claim 1 , wherein the self-energy of the protein backbone is ignored. 
     
     
         31 . A method according to  claim 1 , wherein the ECO formed in step (E) further includes a description of the pattern in terms of the number of pattern residue positions, the location of each pattern residue position within the protein structure, the selection of an amino acid residue type at each position and the conformation of each pattern residue. 
     
     
         32 . A method according to  claim 1 , wherein the ECO further includes a description of or a reference to the protein structure received in step (A). 
     
     
         33 . A method according to  claim 1 , wherein the ECO further includes a description of or a reference to the modeled residue positions defined in step (A). 
     
     
         34 . A method according to  claim 1 , wherein the ECO further includes a description of or a reference to the amino acid reference sequence defined in step (A). 
     
     
         35 . A method according to  claim 1 , wherein the ECO further includes a description of or a reference to the structure optimization method used in step (C). 
     
     
         36 . A method according to  claim 1 , wherein the ECO further includes a description of or a reference to the rotamer library used to assign the conformation of the pattern residues in accordance with step (B) and/or to define the allowed conformational states of the modeled, non-pattern residues in accordance with step (C). 
     
     
         37 . A method according to  claim 1 , wherein the ECO further includes a description of or a reference to the cost function used in step (C). 
     
     
         38 . A method according to  claim 1 , wherein the ECO further includes a description of or a reference to the substituted and optimized protein structure obtained in step (C). 
     
     
         39 . A method according to  claim 1 , wherein the ECO further includes additional information derived from either the reference structure of step (A) or the substituted and optimized structure of step (C), such as the local secondary structure of each modeled residue and/or its accessible surface area. 
     
     
         40 . A method according to  claim 1 , wherein steps (B) to (D) are repeated for different patterns defined in step (B). 
     
     
         41 . A method according to  claim 40 , wherein the patterns are single residue patterns located at one single residue position selected from the set of modeled residue positions of step (A) and are formed by systematically considering all possible combinations of amino acid residue types in different conformations, and wherein the amino acid residue types are selected from a pre-defined list of residue types and the said conformations are selected from a pre-defined list of conformations. 
     
     
         42 . A method according to  claim 40 , wherein the patterns are double residue patterns located at two different single residue position selected from the set of modeled residue positions of step (A) and are formed by systematically considering, for the pair of selected residue positions, all possible combinations of amino acid residue types in different conformations and wherein the amino acid residue types are selected from a pre-defined list of residue types and the said conformations are selected from a pre-defined list of conformations. 
     
     
         43 . A method according to  claim 41 , being repeated until each single residue position of the set of modeled residue positions of step (A) has been selected as a pattern residue position. 
     
     
         44 . A method according to  claim 42 , being repeated until each possible pair of residue position of the set of modeled residue positions of step (A) has been selected as a pair of pattern residue positions. 
     
     
         45 . A method according to  claim 44 , wherein the number of considered double residue patterns is restricted in accordance with a method taking into account a measure of the distance between two residue positions in the reference structure of step (A). 
     
     
         46 . A method according to  claim 45 , wherein the number of considered double residue patterns is restricted in accordance with a method further taking into account a measure of the size of the amino acid residue types placed at each of both pattern residue positions. 
     
     
         47 . A method according to  claim 1 , wherein the quantitative measure representing the energetic compatibility of a pattern within the context of a substituted and adapted reference structure is further transformed. 
     
     
         48 . A method according to  claim 47 , wherein the said transformation is effected by means of a linear, logarithmic or exponential function. 
     
     
         49 . A method according to  claim 40 , wherein the ECO's are further grouped into GECO's and wherein this grouping operation occurs on the basis of the quantitative measure representing the energetic compatibility of different patterns being located at the same residue position(s) and assuming the same residue type(s). 
     
     
         50 . A fold recognition method to identify a potential structural relationship between a particular target amino acid sequence and one or more protein three-dimensional structures, the said protein three-dimensional structures being analyzed by a method according to  claim 1 . 
     
     
         51 . An inverse folding method to identify a potential structural relationship between a particular protein three-dimensional structure and one or more known amino acid sequences, wherein the said protein three-dimensional structure is analyzed by a method according to  claim 1 . 
     
     
         52 . A protein design method to identify or generate amino acid sequences which are energetically compatible with a particular protein three-dimensional structure, wherein the said protein three-dimensional structure is analyzed by a method according to  claim 1 . 
     
     
         53 . A method according to  claim 1 , wherein the cost function includes an energetic contribution accounting for main-chain flexibility. 
     
     
         54 . A method according to  claim 1 , wherein the energy function includes an energetic contribution for solvation effects which is calculated by a method including the assignment of a set of energetic solvation terms to each residue type depending on the degree of solvent exposure of its respective rotamers at the considered residue positions in the protein structure. 
     
     
         55 . A method according to  claim 1 , wherein the energy function includes an energetic contribution for solvation effects which is a type-dependent, topology-specific solvation (TTS) method including establishing a set of energetic parameters for each of different classes of solvent exposure, and wherein the classes of solvent exposure correspond with buried, semi-buried and solvent-exposed rotamers, respectively. 
     
     
         56 . A method according to  claim 55 , wherein the table of TTS values is established by:
 first considering a set of a number of well-resolved protein structures and assigning to each residue position of this set a unique index i,   then considering, at each residue position i, all residue types a in all rotameric states r, thereby forming the set of all possible position-type-rotamer combinations i r   a ,   defining by i r   w  the WT residue type at position i in a rotameric state r as observed in the protein comprising position i,   assigning to each i r   a  a solvent exposure class C(i r   a ),   calculating for each i r   a  the value E pot (i r   a ) as the total potential energy the rotamer i r   a  experiences within the fixed context of the protein structure comprising i,   defining a function ΔE(t) according to equation (2)
   Δ E ( i )=min r ( E   pot ( i   r   w )+ E   TTS ( w,C ( i   r   w )))−min a min r ( E   pot ( i   r   a )+ E   TTS ( a,C ( i   r   a )))  (eq.2)
 
   ΔE(i) being the difference in energy of the WT residue type in its best possible   rotamer and the energetically most favorable residue type at residue position i also in its best possible rotamer, given a set of TTS parameters, and   applying a suitable parameter optimization method to maximize S by varying E TTS (T,C) parameters.   
     
     
         57 . A type-dependent, topology-specific solvation method for the assignment of a set of energetic solvation terms to a set of residue types, depending on the degree of solvent exposure of their respective rotamers at the considered residue positions in a protein structure, wherein:
 first, each residue type at a given residue position in a specific rotameric state is substituted into the protein structure and its accessible surface area (ASA) is calculated,   next, a class assignment occurs on the basis of the percentage ASA of the residue side chain in the protein structure compared to the maximal ASA of the same side chain being shielded from the solvent only by its own main-chain atoms, and   finally an appropriate type- and topology-specific energy, E TTS (T,C), for the considered pattern element is retrieved from a table of TS values, using the pattern type (T) and class (C) indices.   
     
     
         58 . A method according to  claim 57 , wherein the table of TTS values is established by:
 first considering a set of a number of well-resolved protein structures and assigning to each residue position of this set a unique index i,   then considering, at each residue position i, all residue types a in all rotameric states r, thereby forming the set of all possible position-type-rotamer combinations i r   a ,   defining by i r   w  the WT residue type at position i in a rotameric state r as observed in the protein comprising position i,   assigning to each i r   a  a solvent exposure class C(i r   a ),   calculating for each i r   a  the value E pot (i r   a ) as the total potential energy the rotamer i r   a  experiences within the fixed context of the protein structure comprising i,   defining a function ΔE(i) according to equation (2)
   Δ E ( i )=min r ( E   pot ( i   r   w )+ E   TTS ( w,C ( i   r   w )))−min a min r ( E   pot ( i   r   a )+ E   TTS ( a,C ( i   r   a )))  (eq.2)
 
   ΔE(i) being the difference in energy of the WT residue type in its best possible   rotamer and the energetically most favorable residue type at residue position i also in its best possible rotamer, given a set of TTS parameters, and   applying a suitable parameter optimization method to maximize S by varying E TTS (T,C) parameters.   
     
     
         59 . A nucleic acid sequence encoding a protein sequence analyzed by a method according to  claim 1 . 
     
     
         60 . An expression vector comprising the nucleic acid sequence of  claim 59 . 
     
     
         61 . A host cell comprising the nucleic acid sequence of  claim 59 . 
     
     
         62 . A pharmaceutical composition comprising a therapeutically effective amount of a protein sequence analyzed by a method according to  claim 1  and a pharmaceutically acceptable carrier. 
     
     
         63 . A method of treating a disease in a mammal, comprising administering a pharmaceutical composition according to  claim 62  to said mammal in need thereof. 
     
     
         64 . An ECO obtainable by a method according to  claim 1 . 
     
     
         65 . A database in the form of a data structure comprising a set of ECO's obtainable by a method according to  claim 1 . 
     
     
         66 . A computing device for computing a type-dependent, topology-specific solvation for the assignment of a set of energetic solvation terms to a set of residue types, depending on the degree of solvent exposure of their respective rotamers at the considered residue positions in a protein structure, comprising:
 Means for substituting each residue type at a given residue position in a specific rotameric state into the protein structure and its accessible surface area (ASA) is calculated,   Means for a class assigning on the basis of the percentage ASA of the residue side chain in the protein structure compared to the maximal ASA of the same side chain being shielded from the solvent only by its own main-chain atoms, and   Means for retrieving an appropriate type- and topology-specific energy, E TTS (T,C), for the considered pattern element from a table of TTS values in the form of a data structure, using the pattern type (T) and class (C) indices.   
     
     
         67 . A computer program product to be utilized with a computer system having a memory and a processor for computing a type-dependent, topology-specific solvation for the assignment of a set of energetic solvation terms to a set of residue types, depending on the degree of solvent exposure of their respective rotamers at the considered residue positions in a protein structure, comprising:
 Instruction means for substituting each residue type at a given residue position in a specific rotameric state into the protein structure and its accessible surface area (ASA) is calculated,   Instruction means for a class assigning on the basis of the percentage ASA of the residue side chain in the protein structure compared to the maximal ASA of the same side chain being shielded from the solvent only by its own main-chain atoms, and   Instruction means for retrieving an appropriate type- and topology-specific energy, E TTS (T,C), for the considered pattern element from a table of TTS values in the form of a data structure, using the pattern type (T) and class (C) indices.   
     
     
         68 . A computer readable data carrier comprising an executable computer program product in accordance with  claim 67 . 
     
     
         69 . A computer readable data carrier comprising an executable computer program product for executing the method of  claim 1 . 
     
     
         70 . A method comprising the steps of
 inputting a description of a least a protein reference structure at a near location;   transmitting the description to a remote processing engine running a computer program for carrying out any of the methods of  claim 1 ;   receiving at a near location from the remote processing engine an output of a method in accordance with  claim 1 .