- Email: [email protected]
P-A2-16 Coding and classifying primary and tertiary structures of proteins
A2 3D structures: classification, evolution and prediction P-A2-14 PREDICTION OF HELIX CONFIGURATION BY CONTINUUM THEORY WITHOUT ASWMPTION OF 2D STRUCTURAL TEMPLATE HIROKAWA T, SUWA M, MiTAKU S, Department of Biotechnology, Tokyo A & T, Tokyo(Japan)
P-A2-13 QUANTUM-MECHANICAL DERIVATION OF FORCES IN PROTEIN SIMULATIONS CARLSSON, AE. Departmentof Physics, WashingtonUniversity,St. Louis(USA) Purpose: The purposeof this work is to improve the reliability of protein simulations by developing a quantum-mechanically based formalism for calculating the forces.
Purpose: As the second half of a software system for 3D structure prediction of membrane proteins, we have developed a method to determine the positioning, the rotation and the tilting of transmembrane helices, assuming that the polar interactions are essential for the 3D structure formation. Methods: The first stage of the 3D-structure prediction is the search of an energy ground minimum, assuming a triangle lattice and the interhelix binding by polar interactions. In the secondstage. the lattice itself is deformed around the bestco-figuration in the first stage, leading to the final positioning, orientation together with the tilt angle of helices. Results: Analyzing the amino acid sequence of bacteriorhodopsin, almost the same configuration as the experimental result by Henderson et al, (J. Mol. Biol., 1990). Conclusions: All structural space of bacteriorhodopsincould be searched,adopting the continuum theory. Due to the presentwork together with the accompanying paper, it has become possibleto predict the 3D-structure of a membrane protein with only the information of its amino acid sequence.
Methods: The electronic contribution to the forces is calculatedwithin a tight-binding model of the electronic structure, using the “bondorbital” approximation and perturbation theory. Results: The angular and torsional forces display simple angular dependenceswith coefficients given in terms of the electronic couplings in the tight-binding model, including out-of-plane energiesand torsional terms. Conclusions: A formalism has been developed for calculating the functional form of a part of the interatomic forces in proteins from a quantum-mechanicalbasis. The next step should be the development of a practical force field by suitable parametrizationof these functional forms.
P-A2-15 DETE~ATiON QF ~ BELHZAL RmfONS BY PHYSICOCI-IEMKJAL A@PROACH MITAKU S, SBAH B-C. Department of Biotechnology, Tokyo A & T, Tokyo(Japan)
P-A2- 16 CODING AND CLASSIFYING PRIMARY AND TERTIARY STRUCTURES OF PROTEINS. SJRABELLA, P. and COLOSIMO, A. Dept.of Bicchemiiuy.Univ.of Rome“La Sapieuza” (I) Email: [email protected] Purpose: A new coding rocedureto optimize the automaticclass&ation 0%thenrimarvstruehcresof
Purpose: As the first half of a software systemfor 3D structure prediction of membrane proteins, we have developed a new physicochemical method to predict transmembranehelical regions from amino acid sequence. Methods: We have taken four factors into account: (1) hydrophobic@ of segments,(2) polar segments near the hydrophobic segments, (3) change in hydrophobicity due to the masking of polar groups of Trp, Tyr and so on, and (4) stabilizationof helices through direct binding with neighboring helices. Results: Almost 100% of transmembmne helices were truly nredicted for membrane oroteins whose 3D-struciu~eis already known, Thi true negative nrediction was better than 95% for soluble oroteins in SCOPdatabase,the size of which were;maller than 800 residues. Furthermore, transmembrane regionscould be classifiedinto two kinds of helices by physicochemical consideration: structural transmembranehelicesand functional ones. Con&raions: The hydrophobicityalone of segments was not sufficient but the four factors of the structural stability were enough to determine the transmembranehelices.
with the shnilfi&iesin’the te&ary Methods: The rimary $uchues of proteins have beentranformel Into numericalvectors(or profiles) associatedwith one (or more than one) physicochemicalpro rty of the aminoacidicresidues.A Kohonen’sSep”f OrganixingAlgorithmhasbeenused to classifythe numericalprofilesin conjunctionwith: a) an automaticprocedureof reductionto the same lenathof the numericalvectorswith mtnbnumlossof inf&mation; b) a R&&pal ComponentAnaIysisto get rid of the redundant information; c) a new algorithm for the objective classification of the tertiary structures based upon tht tridimensional coordinates. Results: The automaticclassificationbasedon the primary structures and t&e results obtained by completely independent approaches provide substantiallyidentical resultsunder dl the explored
38
P-A2-13 QUANTUM-MECHANICAL DERIVATION OF FORCES IN PROTEIN SIMULATIONS CARLSSON, AE. Departmentof Physics, WashingtonUniversity,St. Louis(USA) Purpose: The purposeof this work is to improve the reliability of protein simulations by developing a quantum-mechanically based formalism for calculating the forces.
Purpose: As the second half of a software system for 3D structure prediction of membrane proteins, we have developed a method to determine the positioning, the rotation and the tilting of transmembrane helices, assuming that the polar interactions are essential for the 3D structure formation. Methods: The first stage of the 3D-structure prediction is the search of an energy ground minimum, assuming a triangle lattice and the interhelix binding by polar interactions. In the secondstage. the lattice itself is deformed around the bestco-figuration in the first stage, leading to the final positioning, orientation together with the tilt angle of helices. Results: Analyzing the amino acid sequence of bacteriorhodopsin, almost the same configuration as the experimental result by Henderson et al, (J. Mol. Biol., 1990). Conclusions: All structural space of bacteriorhodopsincould be searched,adopting the continuum theory. Due to the presentwork together with the accompanying paper, it has become possibleto predict the 3D-structure of a membrane protein with only the information of its amino acid sequence.
Methods: The electronic contribution to the forces is calculatedwithin a tight-binding model of the electronic structure, using the “bondorbital” approximation and perturbation theory. Results: The angular and torsional forces display simple angular dependenceswith coefficients given in terms of the electronic couplings in the tight-binding model, including out-of-plane energiesand torsional terms. Conclusions: A formalism has been developed for calculating the functional form of a part of the interatomic forces in proteins from a quantum-mechanicalbasis. The next step should be the development of a practical force field by suitable parametrizationof these functional forms.
P-A2-15 DETE~ATiON QF ~ BELHZAL RmfONS BY PHYSICOCI-IEMKJAL A@PROACH MITAKU S, SBAH B-C. Department of Biotechnology, Tokyo A & T, Tokyo(Japan)
P-A2- 16 CODING AND CLASSIFYING PRIMARY AND TERTIARY STRUCTURES OF PROTEINS. SJRABELLA, P. and COLOSIMO, A. Dept.of Bicchemiiuy.Univ.of Rome“La Sapieuza” (I) Email: [email protected] Purpose: A new coding rocedureto optimize the automaticclass&ation 0%thenrimarvstruehcresof
Purpose: As the first half of a software systemfor 3D structure prediction of membrane proteins, we have developed a new physicochemical method to predict transmembranehelical regions from amino acid sequence. Methods: We have taken four factors into account: (1) hydrophobic@ of segments,(2) polar segments near the hydrophobic segments, (3) change in hydrophobicity due to the masking of polar groups of Trp, Tyr and so on, and (4) stabilizationof helices through direct binding with neighboring helices. Results: Almost 100% of transmembmne helices were truly nredicted for membrane oroteins whose 3D-struciu~eis already known, Thi true negative nrediction was better than 95% for soluble oroteins in SCOPdatabase,the size of which were;maller than 800 residues. Furthermore, transmembrane regionscould be classifiedinto two kinds of helices by physicochemical consideration: structural transmembranehelicesand functional ones. Con&raions: The hydrophobicityalone of segments was not sufficient but the four factors of the structural stability were enough to determine the transmembranehelices.
with the shnilfi&iesin’the te&ary Methods: The rimary $uchues of proteins have beentranformel Into numericalvectors(or profiles) associatedwith one (or more than one) physicochemicalpro rty of the aminoacidicresidues.A Kohonen’sSep”f OrganixingAlgorithmhasbeenused to classifythe numericalprofilesin conjunctionwith: a) an automaticprocedureof reductionto the same lenathof the numericalvectorswith mtnbnumlossof inf&mation; b) a R&&pal ComponentAnaIysisto get rid of the redundant information; c) a new algorithm for the objective classification of the tertiary structures based upon tht tridimensional coordinates. Results: The automaticclassificationbasedon the primary structures and t&e results obtained by completely independent approaches provide substantiallyidentical resultsunder dl the explored
38