CN109411013A - A kind of group's Advances in protein structure prediction based on the specific Mutation Strategy of individual - Google Patents

Abstract

A kind of group's Advances in protein structure prediction based on the specific Mutation Strategy of individual, under differential evolution algorithm frame, for each conformation, it is calculated at a distance from all conformations other in current population, and determines the state of conformation individual according to the search condition factor of average distance, maximum distance and minimum distance calculation conformation；If it is larger that conformation is in local search shape probability of state, some individuals adjacent thereto are selected, and select the optimum individual in these individuals to instruct the mutation process of conformation；If conformation is in global acquisition mode, conformation is randomly choosed from entire population to make a variation；To which specific Mutation Strategy be arranged to each conformation, achievees the effect that improve sampling efficiency and maintain population diversity.The present invention provides a kind of search efficiency and the higher group's Advances in protein structure prediction based on the specific Mutation Strategy of individual of precision of prediction.

Description

A kind of group's Advances in protein structure prediction based on the specific Mutation Strategy of individual

Technical field

The present invention relates to a kind of biological information, intelligent optimization, computer application field more particularly to a kind of bases In group's Advances in protein structure prediction of the specific Mutation Strategy of individual.

Background technique

Nineteen ninety, the U.S. starts the Human Genome Project, and announces completion in 2003.For over ten years, Ren Leiji Because a group plan deepens constantly the mankind to the understanding of itself and disease, depth is brought to medicine, mathematics, biology and computer science It is remote to influence.However up to now, former US President Clinton discribed blueprint at that time is not yet presented: " it is right thoroughly to change us The diagnosis of most diseases prevents and treats means ".To find out its cause, being the amino acid that genome depicts only protein Sequence (i.e. prlmary structure of protein), and protein is only folded into specific three-dimensional structure (i.e. tertiary structure) could generate it Specific biological function.Relative to the first genetic code, i.e. DNA translates to albumen with the codon that three nucleotide are one group The amino acid sequence of matter, between protein sequence primary structure and its tertiary structure corresponding relationship (i.e. the second genetic code or Code of folding) it is still referred to as unsolved mystery.For opposing proteins fold, protein structure prediction has stronger practicability, only There is the three-dimensional structure for obtaining protein, could really realize gene diagnosis, and be finally reached gene therapy purpose.

Currently, the experimental method of measurement protein three-dimensional structure includes X-ray crystal diffraction, multi-dimensional nmr (NMR) With electron cryo-microscopy etc..X-ray crystal diffraction is current measurement protein structure most efficient method, and precision achieved is other What method cannot compare, major defect is that protein crystal is difficult to cultivate and the period of crystal structure determination is longer；NMR method The conformation of protein in the solution can directly be measured, but it is big to the requirement of sample, purity requirement is high, can only measure at present Small protein.Secondly, these experimental determining methods are expensive, the three-dimensional structure for measuring a protein needs hundreds of thousands Dollar, however, the primary amino acid sequences of one protein of measurement only need 1000 dollars or so, so as to cause protein sequence and Wide gap between three-dimensional structure measurement is increasing.Therefore, such as how computer is tool, with algorithm appropriate, from amino Acid sequence, which sets out, directly predicts the three-dimensional structure of protein, becomes a kind of important research topic in current biological informatics.

Conformational space optimization (or sampling) method be it is current restrict protein structure ab initio prediction precision most critical because One of element.Differential evolution algorithm (Differential Evolution, DE) is as algorithm most powerful in evolution algorithm A kind of randomness algorithm that Price and Storn is proposed in nineteen ninety-five.DE algorithm is since structure is simple, fast convergence rate, robustness The advantages that strong, has in protein conformation space optimization field and is widely applied.Sudha etc. proposes a kind of based on local policy Differential evolution Advances in protein structure prediction；Custodio etc. proposes a kind of group albumen of the local based on similarity agent model Matter structure prediction prediction technique；Shehu research group is based on DE algorithm, proposes a series of effective protein conformation space optimizations Method, such as multiple dimensioned HYBRID EVOLUTIONARY ALGORITHMS HEA and multiple target conformational space optimization method MOEA.However, in DE algorithm, no With Mutation Strategy have a different advantages, such as the global detectivities of some strategies are stronger, and some tactful local search energy Power is stronger.In addition, during evolution, preferably conformation individual and poor conformation individual plays different roles.It is more excellent Individual be responsible for instructing the direction of search of algorithm, and poor individual be responsible for maintain population diversity.Because entirely evolving A kind of conformation Mutation Strategy is used only in the process, can not only reduce the search efficiency of algorithm, but also algorithm is easily trapped into part most It is excellent, and influence final precision of prediction.

Therefore, existing group's Advances in protein structure prediction in terms of precision of prediction and search efficiency there is defect, It needs to improve.

Summary of the invention

In order to overcome the precision of prediction and the lower deficiency of search efficiency of existing group's Advances in protein structure prediction, this hair It is bright to propose a kind of precision of prediction and the higher group's protein structure prediction based on the specific Mutation Strategy of individual of search efficiency Method.

The technical solution adopted by the present invention to solve the technical problems is:

A kind of group's Advances in protein structure prediction based on the specific Mutation Strategy of individual, the method includes following steps It is rapid:

1) sequence information of testing protein is inputted, and from ROBETTA server (http://www.robetta.org/) On obtain fragment library；

2) parameter setting: setting population scale NP, crossover probability CR, fragment length l, temperature factor KT, neighbouring number of individuals Measure N, maximum number of iterations G_max, and initialize the number of iterations g=0；

3) segment assembling is randomly choosed from the corresponding fragment library in each residue position generates initial configurations population P={ C₁, C₂,...,C_NP, wherein C_i, i={ 1,2 ..., NP } is i-th of conformation individual in population P；

4) energy value of each conformation in current population is calculated according to Rosetta socre3 energy function；

5) to each conformation C in population_i, i ∈ 1,2 ..., and NP } it performs the following operations:

5.1) by conformation C_iRegard target conformation as, is calculated between target conformation and other NP-1 conformation according to carbon alpha atom Euclidean distance, the distance between target conformation and j-th of conformation are denoted as d_j；

5.2) NP-1 distance d are calculated_j, j=1,2 ..., the average value d of NP-1_ave, and by this NP-1 distance in most Big distance is denoted as d_max, minimum range is denoted as d_min；

5.3) the search condition factor of target conformation is calculated

5.4) random to generate a decimal between 0 and 1IfThen perform the following operations:

5.4.1 the N number of conformation nearest with target conformation distance) is selected, and selects the conformation of minimum energy in this N number of conformation C_pbest, and two different and and C are randomly selected from this N number of neighbouring conformation_pbestDifferent conformation C_aAnd C_b；

5.4.2) random to generate a random integers R ' between 1 and 2, if R '=1, respectively from conformation C_aAnd C_bIn It randomly chooses the segment that the different length in residue position is l and replaces conformation C_pbestThe segment of middle corresponding position generates variation structure As C_mutant；

5.4.3) if R '=2, respectively from conformation C_pbestAnd C_bThe different length in one residue position of middle random selection is l Segment replace conformation C_aThe segment of middle corresponding position generates variation conformation C_mutant；

If 5.5)Then perform the following operations:

5.5.1) from current population randomly choose three it is different and with target conformation also different conformation C_c、C_d And C_e；

5.5.2) respectively from C_c、C_dAnd C_eThe segment that the different length in one residue position of middle random selection is l replaces target structure As C_iThe segment of middle corresponding position generates variation conformation C_mutant；

5.6) random to generate a decimal R between 0 and 1, if R < CR, from conformation C_iIn randomly select a length For the segment replacement variation conformation C of l_mutantThe segment of middle corresponding position, and a random fragment assembling is carried out, to generate survey Try conformation C_trial；Otherwise variation conformation is directly subjected to a random fragment assembling and generates test conformation C_trial；

5.7) test conformation C is calculated according to Rosetta score3 energy function_trialEnergy value；

If 5.8) C_trialEnergy value be less than C_iEnergy value, then C_trialReplace C_i, otherwise according to Boltzmann probabilityUse C_trialReplace C_i, wherein Δ E is C_trialEnergy value and C_iEnergy value miss absolute value of the difference；

6) g=g+1, if g > G_max, then the final pre- geodesic structure of conformation conduct of minimum energy is exported, otherwise return step 5)。

Technical concept of the invention are as follows: under differential evolution algorithm frame, for each conformation, calculate itself and current population In other all conformations distance, and according to the search condition factor of average distance, maximum distance and minimum distance calculation conformation To determine the state of conformation individual；If it is larger that conformation is in local search shape probability of state, portion adjacent thereto is selected Divide individual, and selects the optimum individual in these individuals to instruct the mutation process of conformation；If conformation is in global detection State randomly chooses conformation from entire population then to make a variation；To which specific Mutation Strategy be arranged to each conformation, reach To the effect for improving sampling efficiency and maintenance population diversity.The present invention provides a kind of search efficiency and precision of prediction is higher Group's Advances in protein structure prediction based on the specific Mutation Strategy of individual.

Beneficial effects of the present invention are shown: its search condition is measured at a distance from other conformations according to each conformation, It designs different Mutation Strategies for the conformation of different conditions and instructs stable conformation, to reach raising and maintain population multiplicity Property effect, and then improve precision of prediction.

Detailed description of the invention

Fig. 1 is that group's Advances in protein structure prediction based on the specific Mutation Strategy of individual carries out structure to protein 1BQ9 Conformation when prediction updates schematic diagram.

Fig. 2 is that group's Advances in protein structure prediction based on the specific Mutation Strategy of individual carries out structure to protein 1BQ9 The conformation distribution map obtained when prediction.

Fig. 3 is that group's Advances in protein structure prediction based on the specific Mutation Strategy of individual carries out structure to protein 1BQ9 Predict obtained tomograph.

Specific embodiment

The invention will be further described below in conjunction with the accompanying drawings.

Referring to Fig.1~Fig. 3, a kind of group's Advances in protein structure prediction based on the specific Mutation Strategy of individual, including with Lower step:

1) sequence information of testing protein is inputted, and from ROBETTA server (http://www.robetta.org/) On obtain fragment library；

2) parameter setting: setting population scale NP, crossover probability CR, fragment length l, temperature factor KT, neighbouring number of individuals Measure N, maximum number of iterations G_max, and initialize the number of iterations g=0；

3) segment assembling is randomly choosed from the corresponding fragment library in each residue position generates initial configurations population P={ C₁, C₂,...,C_NP, wherein C_i, i={ 1,2 ..., NP } is i-th of conformation individual in population P；

4) energy value of each conformation in current population is calculated according to Rosetta socre3 energy function；

5) to each conformation C in population_i, i ∈ 1,2 ..., and NP } it performs the following operations:

5.1) by conformation C_iRegard target conformation as, is calculated between target conformation and other NP-1 conformation according to carbon alpha atom Euclidean distance, the distance between target conformation and j-th of conformation are denoted as d_j；

5.2) NP-1 distance d are calculated_j, j=1,2 ..., the average value d of NP-1_ave, and by this NP-1 distance in most Big distance is denoted as d_max, minimum range is denoted as d_min；

5.3) the search condition factor of target conformation is calculated

5.4) random to generate a decimal between 0 and 1IfThen perform the following operations:

5.4.1 the N number of conformation nearest with target conformation distance) is selected, and selects the conformation of minimum energy in this N number of conformation C_pbest, and two different and and C are randomly selected from this N number of neighbouring conformation_pbestDifferent conformation C_aAnd C_b；

5.4.2) random to generate a random integers R ' between 1 and 2, if R '=1, respectively from conformation C_aAnd C_bIn It randomly chooses the segment that the different length in residue position is l and replaces conformation C_pbestThe segment of middle corresponding position generates variation structure As C_mutant；

5.4.3) if R '=2, respectively from conformation C_pbestAnd C_bThe different length in one residue position of middle random selection is l Segment replace conformation C_aThe segment of middle corresponding position generates variation conformation C_mutant；

If 5.5)Then perform the following operations:

5.5.1) from current population randomly choose three it is different and with target conformation also different conformation C_c、C_d And C_e；

5.5.2) respectively from C_c、C_dAnd C_eThe segment that the different length in one residue position of middle random selection is l replaces target structure As C_iThe segment of middle corresponding position generates variation conformation C_mutant；

5.6) random to generate a decimal R between 0 and 1, if R < CR, from conformation C_iIn randomly select a length For the segment replacement variation conformation C of l_mutantThe segment of middle corresponding position, and a random fragment assembling is carried out, to generate survey Try conformation C_trial；Otherwise variation conformation is directly subjected to a random fragment assembling and generates test conformation C_trial；

5.7) test conformation C is calculated according to Rosetta score3 energy function_trialEnergy value；

If 5.8) C_trialEnergy value be less than C_iEnergy value, then C_trialReplace C_i, otherwise according to Boltzmann probabilityUse C_trialReplace C_i, wherein Δ E is C_trialEnergy value and C_iEnergy value miss absolute value of the difference；

6) g=g+1, if g > G_max, then the final pre- geodesic structure of conformation conduct of minimum energy is exported, otherwise return step 5)。

The β-pleated sheet protein 1BQ9 that the present embodiment sequence length is 53 is embodiment, and one kind is based on the specific variation plan of individual Group's Advances in protein structure prediction slightly, wherein comprising the steps of:

1) sequence information of testing protein is inputted, and from ROBETTA server (http://www.robetta.org/) On obtain fragment library；

2) parameter setting: setting population scale NP=50, crossover probability CR=0.5, fragment length l=3, temperature factor KT =2, neighbouring individual amount N=NP/5, maximum number of iterations G_max=1000, and initialize the number of iterations g=0；

3) segment assembling is randomly choosed from the corresponding fragment library in each residue position generates initial configurations population P={ C₁, C₂,...,C_NP, wherein C_i, i={ 1,2 ..., NP } is i-th of conformation individual in population P；

4) energy value of each conformation in current population is calculated according to Rosetta socre3 energy function；

5) to each conformation C in population_i, i ∈ 1,2 ..., and NP } it performs the following operations:

5.1) by conformation C_iRegard target conformation as, is calculated between target conformation and other NP-1 conformation according to carbon alpha atom Euclidean distance, the distance between target conformation and j-th of conformation are denoted as d_j；

5.2) NP-1 distance d are calculated_j, j=1,2 ..., the average value d of NP-1_ave, and by this NP-1 distance in most Big distance is denoted as d_max, minimum range is denoted as d_min；

5.3) the search condition factor of target conformation is calculated

5.4) random to generate a decimal between 0 and 1IfThen perform the following operations:

5.4.1 the N number of conformation nearest with target conformation distance) is selected, and selects the conformation of minimum energy in this N number of conformation C_pbest, and two different and and C are randomly selected from this N number of neighbouring conformation_pbestDifferent conformation C_aAnd C_b；

5.4.2) random to generate a random integers R ' between 1 and 2, if R '=1, respectively from conformation C_aAnd C_bIn It randomly chooses the segment that the different length in residue position is l and replaces conformation C_pbestThe segment of middle corresponding position generates variation structure As C_mutant；

5.4.3) if R '=2, respectively from conformation C_pbestAnd C_bThe different length in one residue position of middle random selection is l Segment replace conformation C_aThe segment of middle corresponding position generates variation conformation C_mutant；

If 5.5)Then perform the following operations:

5.5.1) from current population randomly choose three it is different and with target conformation also different conformation C_c、C_d And C_e；

5.5.2) respectively from C_c、C_dAnd C_eThe segment that the different length in one residue position of middle random selection is l replaces target structure As C_iThe segment of middle corresponding position generates variation conformation C_mutant；

5.6) random to generate a decimal R between 0 and 1, if R < CR, from conformation C_iIn randomly select a length For the segment replacement variation conformation C of l_mutantThe segment of middle corresponding position, and a random fragment assembling is carried out, to generate survey Try conformation C_trial；Otherwise variation conformation is directly subjected to a random fragment assembling and generates test conformation C_trial；

5.7) test conformation C is calculated according to Rosetta score3 energy function_trialEnergy value；

If 5.8) C_trialEnergy value be less than C_iEnergy value, then C_trialReplace C_i, otherwise according to Boltzmann probabilityUse C_trialReplace C_i, wherein Δ E is C_trialEnergy value and C_iEnergy value miss absolute value of the difference；

6) g=g+1, if g > G_max, then the final pre- geodesic structure of conformation conduct of minimum energy is exported, otherwise return step 5)。

The β-pleated sheet protein 1BQ9 for being 53 using sequence length has obtained the protein with above method as embodiment Nearly native state conformation, lowest mean square root deviation areAverage root-mean-square deviation isPre- geodesic structure is as shown in Figure 3.

Described above is that the present invention is obtained as example using protein 1BQ9 as a result, and non-limiting implementation model of the invention It encloses, various changes and improvements is done to it under the premise of without departing from range involved by basic content of the present invention, should not exclude at this Except the protection scope of invention.

Claims (1)

1. a kind of group's Advances in protein structure prediction based on the specific Mutation Strategy of individual, it is characterised in that: the method packet Include following steps:

1) sequence information of testing protein is inputted, and obtains fragment library from ROBETTA server；

2) parameter setting: setting population scale NP, crossover probability CR, fragment length l, temperature factor KT, neighbouring individual amount N, Maximum number of iterations G_max, and initialize the number of iterations g=0；

3) segment assembling is randomly choosed from the corresponding fragment library in each residue position generates initial configurations population P={ C₁,C₂,..., C_NP, wherein C_i, i={ 1,2 ..., NP } is i-th of conformation individual in population P；

4) energy value of each conformation in current population is calculated according to Rosetta socre3 energy function；

5) to each conformation C in population_i, i ∈ 1,2 ..., and NP } it performs the following operations:

5.1) by conformation C_iRegard target conformation as, the Euclidean between target conformation and other NP-1 conformation is calculated according to carbon alpha atom The distance between target conformation and j-th of conformation are denoted as d by distance_j；

5.2) NP-1 distance d are calculated_j, j=1,2 ..., the average value d of NP-1_ave, and by this NP-1 distance in it is maximum away from From being denoted as d_max, minimum range is denoted as d_min；

5.3) the search condition factor of target conformation is calculated

5.4) random to generate a decimal between 0 and 1IfThen perform the following operations:

5.4.1 the N number of conformation nearest with target conformation distance) is selected, and selects the conformation of minimum energy in this N number of conformation C_pbest, and two different and and C are randomly selected from this N number of neighbouring conformation_pbestDifferent conformation C_aAnd C_b；

5.4.2) random to generate a random integers R ' between 1 and 2, if R '=1, respectively from conformation C_aAnd C_bIn it is random The length for selecting a residue position different replaces conformation C for the segment of l_pbestThe segment of middle corresponding position generates variation conformation C_mutant；

5.4.3) if R '=2, respectively from conformation C_pbestAnd C_bThe piece that the different length in one residue position of middle random selection is l Section replacement conformation C_aThe segment of middle corresponding position generates variation conformation C_mutant；

If 5.5)Then perform the following operations:

5.5.1) from current population randomly choose three it is different and with target conformation also different conformation C_c、C_dAnd C_e；

5.5.2) respectively from C_c、C_dAnd C_eThe segment that the different length in one residue position of middle random selection is l replaces target conformation C_i The segment of middle corresponding position generates variation conformation C_mutant；

5.6) random to generate a decimal R between 0 and 1, if R < CR, from conformation C_iIn to randomly select length be l Segment replacement variation conformation C_mutantThe segment of middle corresponding position, and a random fragment assembling is carried out, to generate test conformation C_trial；Otherwise variation conformation is directly subjected to a random fragment assembling and generates test conformation C_trial；

5.7) test conformation C is calculated according to Rosetta score3 energy function_trialEnergy value；

If 5.8) C_trialEnergy value be less than C_iEnergy value, then C_trialReplace C_i, otherwise according to Boltzmann probabilityUse C_trialReplace C_i, wherein Δ E is C_trialEnergy value and C_iEnergy value miss absolute value of the difference；

6) g=g+1, if g > G_max, then the final pre- geodesic structure of conformation conduct of minimum energy is exported, otherwise return step 5).