CN102831332A - Interpretation prediction method of transmembrane helix of membrane protein - Google Patents

Interpretation prediction method of transmembrane helix of membrane protein Download PDF

Info

Publication number
CN102831332A
CN102831332A CN2012102616131A CN201210261613A CN102831332A CN 102831332 A CN102831332 A CN 102831332A CN 2012102616131 A CN2012102616131 A CN 2012102616131A CN 201210261613 A CN201210261613 A CN 201210261613A CN 102831332 A CN102831332 A CN 102831332A
Authority
CN
China
Prior art keywords
protein
som
rule
fuzzy
amino acid
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Pending
Application number
CN2012102616131A
Other languages
Chinese (zh)
Inventor
於东军
沈红斌
唐振民
杨静宇
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Nanjing University of Science and Technology
Nanjing University of Science and Technology Changshu Research Institute Co Ltd
Original Assignee
Nanjing University of Science and Technology
Nanjing University of Science and Technology Changshu Research Institute Co Ltd
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by Nanjing University of Science and Technology, Nanjing University of Science and Technology Changshu Research Institute Co Ltd filed Critical Nanjing University of Science and Technology
Priority to CN2012102616131A priority Critical patent/CN102831332A/en
Publication of CN102831332A publication Critical patent/CN102831332A/en
Pending legal-status Critical Current

Links

Images

Landscapes

  • Management, Administration, Business Operations System, And Electronic Commerce (AREA)

Abstract

The invention discloses an interpretation prediction method of transmembrane helix of membrane protein. The method comprises the steps of: firstly, obtaining evolutionary information of protein by PSI-BLAST (Position-specific Iterated BLAST (Basic Local Alignment Search Tool)) program, and extracting features of each amino acid residue by a sliding window technology; then learning the distribution rule of transmembrane helix in a feature space by an SOM (Self-organizing Map), and encoding distribution rule knowledge in a weight vector of the SOM; finally, extracting an interpretation fuzzy rule set by a Wang-Mendel method; and predicting each amino acid residue of the given protein to be predicted by a fuzzy inference technology, and determining whether each amino acid residue belongs to a transmembrane helix segment by a dynamic threshold segmentation technology after obtaining a prediction curve. The method has the advantages that: 1, the distribution rule knowledge of the transmembrane helix is excavated by the learning of the SOM, and the noise of original data is reduced; and 2, the prediction mode of the transmembrane helix of the membrane protein, obtained by the fuzzy rule extraction technology, has very high interpretation performance.

Description

A kind of memebrane protein transbilayer helix Forecasting Methodology of interpretation
Technical field
The present invention relates to memebrane protein sequence transbilayer helix forecasting techniques, particularly a kind of transbilayer helix Forecasting Methodology with high interpretation.
Background technology
Memebrane protein (Transmembrane Protein) is one type of very important protein in biosome, and it all plays important effect for nutriment transportation, intercellular signal transmission and the energy exchange of cell.Simultaneously, memebrane protein also is a lot of pharmaceutically-active target spots, and most typical is the G protein family.There are some researches show that 60% ~ 70% target protein is G protein family member in the medicament research and development.In genomic data, there is 20% ~ 30% gene outcome to be predicted to be memebrane protein, yet regrettably, in PDB (Protein Data Bank) database, has only the transmembrane protein structure about 1% accurately to be measured.Because the hydrophobic property of memebrane protein; Make that the biologicall test of its structure is very difficult: it need combine with biological membrane to form stable native conformation; Be difficult to obtain crystal structure, and measure the most frequently used being to use X ray to carry out crystal diffraction and using nuclear magnetic resonance technique to measure of protein three-dimensional structure.It is all very unfavorable that the special construction of memebrane protein makes that these two kinds of methods implement.Therefore the relevant knowledge of applying biological information science, the membrane structure of striding that the forecasting techniques that uses a computer is studied memebrane protein just seems particularly important, for finding and being familiar with new transmembrane protein and studying its structure and physiological function has great significance.
Had a lot of memebrane protein transbilayer helix forecast models to occur at present, the precision of prediction of transbilayer helix improves just day by day.At present, some memebrane protein transbilayer helix Forecasting Methodologies occurred, typically had: TMHMM (A. Krogh; B. Larsson, G. von Heijne, and E. L. Sonnhammer; " Predicting transmembrane protein topology with a hidden Markov model:application to complete genomes, " J. Mol. Biol., vol. 305; Pp. 567-580,2001.) and PHOBIUS (L. Kall, A. Krogh; And E. L. Sonnhammer, " A combined transmembrane topology and signal peptide prediction method, " J. Mol. Biol.; Vol. 338, pp. 1027-36,2004.); These two kinds of methods are all used hidden Markov model, and (Hidden Markov Model HMM) carries out the prediction of transbilayer helix; Method based on neural network and dynamic programming; Like MEMSAT3 (Improving the accuracy of transmembrane protein topology prediction using evolutionary information. Bioinformatics; 23 (5): 538-544,2007); Based on the method for support vector base, like SVMtm (Z. Yuan, J. S. Mattick; And R. D. Teasdale; " SVMtm:Support vector machines to predict transmembrane segments, " J. Comput. Chem., vol.25; Pp. 632 – 636,2004).
Yet these forecast models of analysis-by-synthesis can be found, what much more more their paid close attention to is degree of accuracy and the generalization ability of pursuing model, and does not consider containing and the interpretability of computation model to domain knowledge well, has ignored the interpretation of model.Computation model the time more similarly is a black box in work, lacks the effective explanation to the inherent mechanism of result of calculation, and the user is difficult to understand the inner link that exists between the forecast model input and output, also hinder with the biologist between communication with exchange.Therefore, under the prerequisite of precision that guarantees forecast model and generalization ability, how effectively strengthening its interpretation, is an urgent demand that vast Bioexperiment researcher proposes.
The method that the present invention proposes adopts the Fuzzy Rule Sets inference technology to design the prediction of memebrane protein transbilayer helix based on the evolution information of memebrane protein, has preferable model interpretation.Used PSI-BLAST program (A. A. Schaffer et al.; " Improving the accuracy of PSI-BLAST protein database searches with composition-based statistics and other refinements; " Nucleic Acids Res.; Vol. 29, pp. 2994 – 3005,2001) extract the evolution information of memebrane protein; At pattern feature regularity of distribution learning phase, used self-organizing map neural network (T. Kohonen, Self-Organization and Associative Memory, 3rd ed. New York:Springer-Verlag, 1989.); In the Rule Extraction stage, used the Learning-from-example technology that Wang-Mendel proposes (L. X. Wang and J. M. Mendel, " Generating fuzzy rules by learning from examples; " IEEE Trans. System., Man, Cybernetics; Vol. 22; No. 6, pp. 1414 – 1427,1992.).
Summary of the invention
The object of the present invention is to provide a kind of memebrane protein transbilayer helix Forecasting Methodology with high interpretation.
Technical scheme of the present invention is: a kind of memebrane protein transbilayer helix Forecasting Methodology of interpretation, and it may further comprise the steps:
The first step: feature extraction converts the amino acid residue in the protein sequence into vector form and representes.For a protein of forming by individual amino acid; Can obtain its ad-hoc location through the PSI-BLAST algorithm and get sub matrix (Position Specific Scoring Matrix; PSSM); This matrix is row 20 row; Earlier this PSSM is carried out standardization line by line, use sliding window technique to obtain the eigenmatrix of each amino acid residue then; Eigenmatrix is averaged by row; Obtain 20 dimensional feature vectors of this residue:
Figure 268887DEST_PATH_IMAGE001
, wherein which residue representes;
Second step: the study of the pattern feature regularity of distribution, (Self-organizing Map SOM), learns the regularity of distribution of protein transbilayer helix sample, and eliminates the original training sample noise in feature space to use self-organizing map neural network.For given training sample set ; Wherein 0 representes the non-film of striding; Film is striden in 1 expression; Use the batch learning algorithm to train SOM, up to the SOM convergence or reach predefined study step number;
The 3rd step: fuzzy rule extracts, and extracts Fuzzy Rule Sets from the weight vector (codebook vectors) of the SOM that trains.Use the Wang-Mendel rule extraction to concentrate from the weight vector of SOM and extract Fuzzy Rule Sets, emphasis solves inconsistency and conflict property in the leaching process;
The 4th step: protein transbilayer helix prediction; For given protein to be predicted, use the Fuzzy Rule Sets of being extracted in the above-mentioned steps three, use fuzzy reasoning method; Transbilayer helix property to wherein amino acid residue is predicted one by one; Obtain prediction curve, use the method for Threshold Segmentation then, confirm whether each residue belongs to the transbilayer helix fragment.
Beneficial effect of the present invention:
The present invention compares with the pre existing survey technology; Its remarkable advantage: (1) interpretation: though the identification of existing memebrane protein transbilayer helix section has reached precision preferably; But generally do not have interpretation; The present invention makes the forecast model of being realized have high interpretation through using the fuzzy rule inference technology; (2) improve predetermined speed and precision: the Wang-Mendel method is extracting rule from the prototype vector of SOM; Rather than from original training sample extracting rule; Effectively reduce regular number, can obtain Fuzzy Rule Sets more closely, improved fuzzy prediction speed and precision effectively.
Description of drawings
Fig. 1 is the process flow diagram of interpretation memebrane protein transbilayer helix Forecasting Methodology.
Fig. 2 is the classification mark figure of the SOM after the training.
Fig. 3 is fuzzy subset's membership function definition figure.
The visable representation of Fig. 4 rule 1.
The visable representation of Fig. 5 rule 2.
The visable representation of Fig. 6 rule 3.
The visable representation of Fig. 7 rule 4.
Fig. 8 protein P16102 strides film tendentiousness curve.
Embodiment
Below in conjunction with accompanying drawing the present invention is further described.
Fig. 1 has provided process flow diagram of the present invention:
At first, use PSI-BLAST to obtain the PSSM matrix of training protein; Secondly, make up training dataset from the PSSM matrix; Then, use SOM that training dataset is learnt; At last, use the Wang-Mendel method from the weight vector of the SOM that trains, to extract Fuzzy Rule Sets.Forecast period for given protein, uses Fuzzy Inference, predicts that successively each residue belongs to the probability of transbilayer helix, uses thresholding method then, judges whether each residue belongs to transbilayer helix.Below, specifically set forth:
The first step; For a protein of forming by
Figure DEST_PATH_IMAGE004
individual amino acid residue; Can obtain ad-hoc location through PSI-BLAST and get sub matrix PSSM (
Figure 965764DEST_PATH_IMAGE005
row 20 row), the information of protein preface is changed into matrix form:
(1)
PSSM is carried out normalization line by line:
(2)
Wherein
Figure DEST_PATH_IMAGE008
(3)
Figure 237663DEST_PATH_IMAGE009
(4)
Using size is the moving window of , extracts the eigenmatrix of each residue:
Figure 60125DEST_PATH_IMAGE011
(5)
Then, calculate the average that each lists, obtain the proper vector of 20 dimensions of corresponding residue:
Figure DEST_PATH_IMAGE012
(6)
Wherein,
Figure DEST_PATH_IMAGE013
(7)
The set of eigenvectors of all proteins residue has constituted training dataset
Figure DEST_PATH_IMAGE014
; Wherein 0 representes the non-film of striding, and film is striden in 1 expression.
Second step: pattern feature regularity of distribution study; Just use
Figure DEST_PATH_IMAGE015
, use the batch learning algorithm and train SOM.In the first?
Figure DEST_PATH_IMAGE016
second batch learning, first calculate the Voronoi? area?
Figure DEST_PATH_IMAGE017
, where?
Figure DEST_PATH_IMAGE018
;?
Figure DEST_PATH_IMAGE019
The vector sum is denoted by?
Figure 134348DEST_PATH_IMAGE020
; Then, use the following iterative formula updating the weights of SOM Vector:
Figure DEST_PATH_IMAGE021
(8)
After the trained, each output node of SOM is marked, output node is divided into two types: 0 and 1, correspondingly respectively non-ly stride the film classification and stride the film classification, as shown in Figure 2.Can find out that from this figure two types distribution has regularity.
The 3rd step: the weight vector of the SOM after the training extracts Fuzzy Rule Sets.If
Figure 333249DEST_PATH_IMAGE022
is the set of the weight vector of training back SOM; Wherein:
Figure DEST_PATH_IMAGE023
is the weight vector of 20 dimensions;
Figure 64444DEST_PATH_IMAGE024
, belonging to classification 0 corresponding to this weight vector still is classification 1.Then, use the Wang-Mendel method from
Figure DEST_PATH_IMAGE025
, to extract Fuzzy Rule Sets:
(1) ambiguity in definition subclass in input and output region.
Let and
Figure DEST_PATH_IMAGE027
were the first
Figure 812137DEST_PATH_IMAGE028
input and output variables of the universe.Definition
Figure DEST_PATH_IMAGE029
individual fuzzy subset and corresponding subordinate function on each domain.Note, can define the fuzzy subset of different numbers on the domain of different variablees, and also can adopt different subordinate functions.This patent is got when realizing.Fig. 3 (a), 3 (b), 3 (c) and 3 (d) provided respectively the 1st, 2,3 and output dimension on fuzzy subset's definition: on each dimension, defined two fuzzy subsets, be respectively " little " and " greatly ", and used the triangle subordinate function.
(2), use the Wang-Mendel method to extract Fuzzy Rule Sets for each weight vector in
Figure DEST_PATH_IMAGE031
.
At first; Corresponding current training sample
Figure 885584DEST_PATH_IMAGE032
selects to have on each dimension the fuzzy subset of maximum membership function value.if
Figure DEST_PATH_IMAGE033
;
Figure 315428DEST_PATH_IMAGE034
;
Figure DEST_PATH_IMAGE068
; ; ;
Figure 557054DEST_PATH_IMAGE036
ties up
Figure 832177DEST_PATH_IMAGE029
the individual fuzzy subset on the domain for being defined in
Figure DEST_PATH_IMAGE037
, and
Figure 537965DEST_PATH_IMAGE038
is the subordinate function of
Figure DEST_PATH_IMAGE039
.So, it is selected to have the fuzzy subset
Figure DEST_PATH_IMAGE041
of maximum membership function value on
Figure 873131DEST_PATH_IMAGE040
dimension:
Figure 602053DEST_PATH_IMAGE042
(9)
Similarly; if
Figure DEST_PATH_IMAGE043
; ;
Figure 164277DEST_PATH_IMAGE068
;
Figure DEST_PATH_IMAGE045
;
Figure 117506DEST_PATH_IMAGE068
;
Figure 991894DEST_PATH_IMAGE046
for being defined in
Figure 763541DEST_PATH_IMAGE029
the individual fuzzy subset on the output domain,
Figure DEST_PATH_IMAGE047
is the subordinate function of .So; Corresponding to output valve , selected fuzzy subset is :
Figure 11486DEST_PATH_IMAGE050
(10)
So far, each dimension goes up selected fuzzy subset's combination and obtains a Fuzzy Rule Sets
IF ?is
Figure 954034DEST_PATH_IMAGE052
?and
Figure DEST_PATH_IMAGE053
?and
Figure 391969DEST_PATH_IMAGE054
?is
Figure 812586DEST_PATH_IMAGE056
?and
Figure DEST_PATH_IMAGE057
?is ?THEN
Figure DEST_PATH_IMAGE059
?is
(3) possesses validity for rule.The availability of rule is the product that each dimension goes up membership function value
(10)
(4) create fuzzy rule base.A rule be can extract from every sample, when sample size is big, redundancy rule and conflict rule occurred possibly.So-called redundancy rule is meant that those have the rule of identical former piece and consequent; Conflict rule is meant that those have identical former piece but the rule of different consequents.Adding one to rule base when regular, whether the rule with the identical former piece of current rule is arranged in the judgment rule storehouse at first; If no, then be added into rule base; If have, then in rule base, only keep the rule of maximum availability.This scheme can effectively solve redundancy rule and conflict rule problem.
The 4th step: protein transbilayer helix prediction; For given protein to be predicted; Use the Fuzzy Rule Sets of being extracted; Use fuzzy reasoning method: to each amino acid residue in this protein, a characteristic of correspondence vector is arranged:
Figure 136885DEST_PATH_IMAGE062
; Then, use fuzzy reasoning to predict:
Use the activity of every rule of product reasoning and calculation
Figure DEST_PATH_IMAGE063
(11)
Reverse gelatinization inference machine in employing center can obtain predicted value (possibility that belongs to transbilayer helix):
Figure 361193DEST_PATH_IMAGE064
(12)
Wherein,
Figure DEST_PATH_IMAGE065
is output fuzzy subset's in
Figure 485007DEST_PATH_IMAGE002
bar rule central value; is predicted value, and is regular number.
At last, the possibility that each amino acid residue is belonged to transbilayer helix is drawn as curve, uses the dynamic threshold cutting techniques can obtain each amino acid residue then and whether belongs to transbilayer helix.
Instance: the memebrane protein database of use comprises 131 memebrane proteins, and wherein the prokaryotes memebrane protein is 92,37 of eukaryotic membrane proteins, and 2 of virus membrane antigens amount to 445 transbilayer helix fragments, and average transmembrane length is 21 residues.Use this patent method, obtain 191 of fuzzy rules altogether.Given 4 typical regular examples below:
Rule 1:
IF?x1?is?F[1,1]?and?x2?is?F[2,1]?and?x3?is?F[3,1]?and?x4?is?F[4,1]?and?x5?is?F[5,1]?and?x6?is?F[6,1]?and?x7?is?F[7,2]?and?x8?is?F[8,2]?and?x9?is?F[9,2]?and?x10?is?F[10,2]?and?x11?is?F[11,1]?and?x12?is?F[12,1]?and?x13?is?F[13,2]?and?x14?is?F[14,2]?and?x15?is?F[15,2]?and?x16?is?F[16,2]?and?x17?is?F[17,2]?and?x18?is?F[18,2]?and?x19?is?F[19,2]?and?x20?is?F[20,2]
THEN?y?is?F[1]?with rule?degree?=?0.002
Rule 2:
IF?x1?is?F[1,1]?and?x2?is?F[2,1]?and?x3?is?F[3,1]?and?x4?is?F[4,1]?and?x5?is?F[5,1]?and?x6?is?F[6,1]?and?x7?is?F[7,2]?and?x8?is?F[8,1]?and?x9?is?F[9,1]?and?x10?is?F[10,2]?and?x11?is?F[11,1]?and?x12?is?F[12,1]?and?x13?is?F[13,1]?and?x14?is?F[14,2]?and?x15?is?F[15,2]?and?x16?is?F[16,2]?and?x17?is?F[17,2]?and?x18?is?F[18,2]?and?x19?is?F[19,2]?and?x20?is?F[20,2]
THEN?y?is?F[1]?with rule?degree?=?0.013
Rule 3:
IF?x1?is?F[1,2]?and?x2?is?F[2,2]?and?x3?is?F[3,2]?and?x4?is?F[4,2]?and?x5?is?F[5,2]?and?x6?is?F[6,2]?and?x7?is?F[7,2]?and?x8?is?F[8,2]?and?x9?is?F[9,2]?and?x10?is?F[10,2]?and?x11?is?F[11,2]?and?x12?is?F[12,1]?and?x13?is?F[13,1]?and?x14?is?F[14,1]?and?x15?is?F[15,1]?and?x16?is?F[16,1]?and?x17?is?F[17,1]?and?x18?is?F[18,1]?and?x19?is?F[19,1]?and?x20?is?F[20,1]
THEN?y?is?F[2]?with rule?degree?=?0.005
Rule 4:
IF?x1?is?F[1,2]?and?x2?is?F[2,2]?and?x3?is?F[3,2]?and?x4?is?F[4,2]?and?x5?is?F[5,2]?and?x6?is?F[6,2]?and?x7?is?F[7,1]?and?x8?is?F[8,1]?and?x9?is?F[9,1]?and?x10?is?F[10,1]?and?x11?is?F[11,2]?and?x12?is?F[12,2]?and?x13?is?F[13,1]?and?x14?is?F[14,1]?and?x15?is?F[15,1]?and?x16?is?F[16,1]?and?x17?is?F[17,1]?and?x18?is?F[18,1]?and?x19?is?F[19,1]?and?x20?is?F[20,1]
THEN?y?is?F[2]?with rule?degree?=?0.029
Rule 1 and 2 is to be used to judge that a residue does not belong to the transbilayer helix fragment, and rule 3 and 4 is to be used to judge that a residue belongs to the transbilayer helix fragment.For explaining that these rules have good interpretation, at first these four rules are converted into the form of figure, meaning as shown in Figure 4.This figure can well interpretative rule, and like the implication of rule 1 and 2 expression be: the IF residue has high water wettability and has low hydrophobicity, and that this residue of THEN does not belong to transbilayer helix; And the implication of rule 3 and 4 expressions is: the IF residue has low water wettability and has high hydrophobicity, and that this residue of THEN belongs to transbilayer helix.The knowledge that these rules are embodied is consistent with human present biology cognition fully.
Transbilayer helix fragment with predicted protein matter P16102 is an example below.The amino acid sequence of protein P16102 is as follows:
>P16102
MSITSVPGVVDAGVLGAQSAAAVRENALLSSSLWVNVALAGIAILVFVYM
GRTIRPGRPRLIWGATLMIPLVSISSYLGLLSGLTVGMIEMPAGHALAGE
MVRSQWGRYLTWALSTPMILLALGLLADVDLGSLFTVIAADIGMCVTGLA
AAMTTSALLFRWAFYAISCAFFVVVLSALVTDWAASASSAGTAEIFDTLR
VLTVVLWLGYPIVWAVGVEGLALVQSVGVTSWAYSVLDVFAKYVFAFILL
RWVANNERTVAVAGQTLGTMSSDD
This protein has 7 transmembrane segments, that is: 27-50,63-82,106-122,134-153,159-179,201-221 and 227-248.
Use this patent method, at first generate the PSSM matrix of this protein; Then, use sliding window technique to extract the characteristic of each residue; With the input of the characteristic of each residue, obtain the tendentiousness that this residue belongs to transbilayer helix as indistinct logic computer; At last, the tendentiousness of all residues in this protein is drawn as curve, as shown in Figure 5; Use Threshold Segmentation technology, the transbilayer helix fragment that can obtain predicting: 29-50,62-86,106-124,132-155,160-181,197-217,225-251.Can find out that from this instance it is fairly good that prediction result and actual value are coincide.
The foregoing description does not limit the present invention in any way, and every employing is equal to the technical scheme that replacement or the mode of equivalent transformation obtain and all drops in protection scope of the present invention.

Claims (3)

1. the memebrane protein transbilayer helix Forecasting Methodology of an interpretation is characterized in that may further comprise the steps:
The first step: feature extraction converts the amino acid residue in the protein sequence into vector form and representes; For a protein of forming by individual amino acid; Obtain its ad-hoc location through the PSI-BLAST algorithm and get sub matrix (Position Specific Scoring Matrix; PSSM); This matrix is
Figure 504234DEST_PATH_IMAGE001
row 20 row; Earlier this PSSM is carried out standardization line by line; Use sliding window technique to obtain the eigenmatrix of each amino acid residue then; Eigenmatrix is averaged by row; Obtain 20 dimensional feature vectors of this residue:
Figure 470922DEST_PATH_IMAGE002
, wherein represent which residue;
Second step: pattern feature regularity of distribution study; Use self-organizing map neural network (Self-organizing Map; SOM); The regularity of distribution of learning sample in feature space, and eliminate the original training sample noise, for given training sample set ; Wherein 0 representes the non-film of striding; Film is striden in 1 expression, uses the batch learning algorithm to train SOM, up to the SOM convergence or reach predefined study step number;
The 3rd step: fuzzy rule extracts, and extracts fuzzy rule from the weight vector (codebook vectors) of the SOM that trains, uses the Wang-Mendel rule extraction to concentrate from the weight vector of SOM and extracts Fuzzy Rule Sets;
The 4th step: protein transbilayer helix prediction; For given protein to be predicted, use the Fuzzy Rule Sets of being extracted in the above-mentioned steps three, use fuzzy reasoning method; Transbilayer helix property to wherein amino acid residue is predicted one by one; Obtain prediction curve, use the method for Threshold Segmentation then, confirm whether each residue belongs to the transbilayer helix fragment.
2. memebrane protein transbilayer helix Forecasting Methodology according to claim 1 is characterized in that the fuzzy subset of the different numbers of definition on the domain of different variablees in the described step 3.
3. memebrane protein transbilayer helix Forecasting Methodology according to claim 1 is characterized in that in the described step 4 adopting the activity of every rule of product reasoning and calculation.
CN2012102616131A 2012-04-16 2012-07-27 Interpretation prediction method of transmembrane helix of membrane protein Pending CN102831332A (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN2012102616131A CN102831332A (en) 2012-04-16 2012-07-27 Interpretation prediction method of transmembrane helix of membrane protein

Applications Claiming Priority (3)

Application Number Priority Date Filing Date Title
CN201210109731.0 2012-04-16
CN201210109731 2012-04-16
CN2012102616131A CN102831332A (en) 2012-04-16 2012-07-27 Interpretation prediction method of transmembrane helix of membrane protein

Publications (1)

Publication Number Publication Date
CN102831332A true CN102831332A (en) 2012-12-19

Family

ID=47334465

Family Applications (1)

Application Number Title Priority Date Filing Date
CN2012102616131A Pending CN102831332A (en) 2012-04-16 2012-07-27 Interpretation prediction method of transmembrane helix of membrane protein

Country Status (1)

Country Link
CN (1) CN102831332A (en)

Cited By (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN103500292A (en) * 2013-09-27 2014-01-08 南京理工大学 Ligand specificity protein-ligand binding area forecasting method
CN104504299A (en) * 2014-12-29 2015-04-08 中国科学院深圳先进技术研究院 Method for predicting action relation between residues of membrane protein
CN104615911A (en) * 2015-01-12 2015-05-13 上海交通大学 Method for predicting membrane protein beta-barrel transmembrane area based on sparse coding and chain training
CN110390995A (en) * 2019-07-01 2019-10-29 上海交通大学 α spiral transmembrane protein topological structure prediction technique and device

Non-Patent Citations (2)

* Cited by examiner, † Cited by third party
Title
DONG-JUN YU 等: "SOMPNN：an efficient non-parametric model for predicting transmembrane helices", 《AMINO ACIDS》 *
DONGJUN YU 等: "SOMRuler： A Novel Interpretable Transmembrane Helices Predictor", 《IEEE TRANSACTIONS ON NANOBIOSCIENCE》 *

Cited By (7)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN103500292A (en) * 2013-09-27 2014-01-08 南京理工大学 Ligand specificity protein-ligand binding area forecasting method
CN104504299A (en) * 2014-12-29 2015-04-08 中国科学院深圳先进技术研究院 Method for predicting action relation between residues of membrane protein
CN104504299B (en) * 2014-12-29 2017-11-03 中国科学院深圳先进技术研究院 The method for predicting the interactively between the residue of memebrane protein
CN104615911A (en) * 2015-01-12 2015-05-13 上海交通大学 Method for predicting membrane protein beta-barrel transmembrane area based on sparse coding and chain training
CN104615911B (en) * 2015-01-12 2017-07-18 上海交通大学 Method based on sparse coding and chain study prediction memebrane protein beta barrel trans-membrane regions
CN110390995A (en) * 2019-07-01 2019-10-29 上海交通大学 α spiral transmembrane protein topological structure prediction technique and device
CN110390995B (en) * 2019-07-01 2022-03-11 上海交通大学 Alpha spiral transmembrane protein topological structure prediction method and device

Similar Documents

Publication Publication Date Title
Ma et al. Learning to iteratively solve routing problems with dual-aspect collaborative transformer
CN113299354B (en) Small molecule representation learning method based on transducer and enhanced interactive MPNN neural network
Baliarsingh et al. Analysis of high-dimensional biomedical data using an evolutionary multi-objective emperor penguin optimizer
Bogdanov et al. Molecular function prediction using neighborhood features
Nayak et al. Hybrid chemical reaction based metaheuristic with fuzzy c-means algorithm for optimal cluster analysis
CN106599611A (en) Marking method and system for protein functions
CN102831332A (en) Interpretation prediction method of transmembrane helix of membrane protein
CN102760209A (en) Transmembrane helix predicting method for nonparametric membrane protein
CN115661550A (en) Graph data class imbalance classification method and device based on generation countermeasure network
Xiong et al. Recursive learning for sparse Markov models
Romero-Zaliz et al. A multiobjective evolutionary conceptual clustering methodology for gene annotation within structural databases: a case of study on the gene ontology database
CN107330550A (en) Space cylindricity assessment method based on double annealing learning aid algorithm
CN116343915B (en) Construction method of biological sequence integrated classifier and biological sequence prediction classification method
Castro et al. Guided generative protein design using regularized transformers
Mondal et al. Network based prediction of protein localisation using diffusion kernel
CN115563312A (en) Medicine-disease-target triple target entity completion method and application
Reyaz-Ahmed et al. Granular decision tree and evolutionary neural SVM for protein secondary structure prediction
Abu Khurma et al. Advanced RIME architecture for global optimization and feature selection
Saini et al. Genetic algorithm for an optimized weighted voting scheme incorporating k-separated bigram transition probabilities to improve protein fold recognition
Li et al. Supervised Reconstruction for High-Dimensional Expensive Multiobjective Optimization
Masoomkhah et al. Deep Learning in Drug Design—Progress, Methods, and Challenges
Zamani Protein secondary structure prediction evaluation and a novel transition site model with new encoding schemes
Yu et al. StackRAM: a cross-species method for identifying RNA N6-methyladenosine sites based on stacked ensemble
Sreelekshmi et al. Dingo optimized fuzzy Cnn technique for efficient protein structure prediction
Liu et al. Effective drug-target affinity prediction via generative active learning

Legal Events

Date Code Title Description
C06 Publication
PB01 Publication
C10 Entry into substantive examination
SE01 Entry into force of request for substantive examination
C02 Deemed withdrawal of patent application after publication (patent law 2001)
WD01 Invention patent application deemed withdrawn after publication

Application publication date: 20121219