CN106528668B - A kind of second order metabolism mass spectrum compound test method based on visual network - Google Patents
A kind of second order metabolism mass spectrum compound test method based on visual network Download PDFInfo
- Publication number
- CN106528668B CN106528668B CN201610925871.3A CN201610925871A CN106528668B CN 106528668 B CN106528668 B CN 106528668B CN 201610925871 A CN201610925871 A CN 201610925871A CN 106528668 B CN106528668 B CN 106528668B
- Authority
- CN
- China
- Prior art keywords
- network
- mass spectrum
- visual
- test method
- compound
- 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.)
- Expired - Fee Related
Links
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/20—Information retrieval; Database structures therefor; File system structures therefor of structured data, e.g. relational data
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
- G06F18/214—Generating training patterns; Bootstrap methods, e.g. bagging or boosting
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/22—Matching criteria, e.g. proximity measures
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/23—Clustering techniques
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
- G06F18/241—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches
- G06F18/2411—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches based on the proximity to a decision surface, e.g. support vector machines
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Data Mining & Analysis (AREA)
- Physics & Mathematics (AREA)
- General Engineering & Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Life Sciences & Earth Sciences (AREA)
- Artificial Intelligence (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Bioinformatics & Computational Biology (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Evolutionary Biology (AREA)
- Evolutionary Computation (AREA)
- Databases & Information Systems (AREA)
- Other Investigation Or Analysis Of Materials By Electrical Means (AREA)
Abstract
The invention discloses a kind of, and the second order based on visual network is metabolized mass spectrum compound test method, wherein, the method is based on existing second order metabolism mass spectrometry database, by carrying out building visual network operation to each compound second order mass spectrometric data, its network global characteristics is extracted as to the input feature vector of its respective compound, and it is trained by SVM, obtain compound test model, and cross validation is carried out to detection model by building Decoy test set, guarantee the reliability and accuracy of detection model, it is set to can be used for the detection of practical second order metabolism mass spectrum compound.Second order metabolism mass spectrum compound test method provided by the invention is realized simple and significantly improves the detection speed and precision of compound.
Description
Technical field
The present invention relates to mass spectrum detection fields more particularly to a kind of second order based on visual network to be metabolized matter
Compose compound test method.
Background technique
Metabolin is the small molecular organic compounds general name that metabolic process is completed in organism, contains physiology shape abundant
State information.Metabolism group can effectively disclose the actual mechanism of physiological phenomenon behind based on the total system Journal of Sex Research to metabolin,
And more fully show the dynamical state of life entity.Therefore more and more attention are obtained, is widely used in many sections
Grind in practical field.
Mass spectral analysis (Mass Spectrometry, MS) is mostly important one of the research tool of metabolism group, existing
Metabolic compounds matching process is mostly based on single order metabolism mass spectrometric data, and process includes three key steps: 1) peak value is examined
It surveys, by the preprocessed elimination noise jamming of raw mass spectrum, obtains effective peak.Common Preprocessing Algorithm includes normalization
(Standardization), PCA albefaction, ZCA albefaction etc.；2) peak value annotation (Annotation) determines target peak (group)
Corresponding specific metabolite type, this process are often accomplished manually by experimenter；3) compound determines, with metabolism
Based on compound mass spectrometry database, the mirror point of compound is carried out by comparing the M/Z value and relative concentration of each spectrum, is commonly used
Metabolic compounds mass spectrometry database include small molecule metabolic pathway database (SMPDB), mankind's metabolin database (HMDB)
Deng.
However this traditional matching algorithm based on single order metabolism mass spectrometric data is often difficult to cope with metabolism group feature
The data characteristics of high-dimensional, small sample, strong noise, its shortcoming is that:
First, existing mass spectrometry value matching process is carrying out the compound matched incipient stage, needs to search for all possibility
This mass spectrographic molecular formula is matched, and same mass spectrometric data may correspond to multiple compound molecule formulas.Research shows that only carrying out one
The matching of rank peak value, the corresponding compound molecule formula of same mass spectrum may carry out second order mass spectrum matching up to more than 100,
Corresponding molecular formula can drop within 5；
Second, existing compound test method depends on extremely accurate M/Z value, and the annotation of peak value needs professional
Depth participate in, required time and cost are all higher.The matching of second order mass spectrum compound is carried out by single M/Z value simultaneously, such as
Fruit compound structure is complex, and testing result precision is difficult to meet requirement of experiment；
Third, existing generation network struction algorithm are to need to consider each compound based on Structure Matching Algorithm mostly
Structure feature, this process is complex, needs a large amount of manual interventions, but the cutting threshold value being manually set is theoretically unsound, and leads
Cause final result unsatisfactory, the time required to calculating and cost is all higher.
Therefore, the existing technology needs to be improved and developed.
Summary of the invention
In view of above-mentioned deficiencies of the prior art, the purpose of the present invention is to provide a kind of the second order generation based on visual network
Thank to mass spectrum compound test method, it is intended to solve the existing antimetabolic compoundis object detecting method people based on single order metabolism mass spectrometric data
Work intervention is more, testing result precision is poor and calculates required time and all higher problem of cost.
Technical scheme is as follows:
A kind of second order metabolism mass spectrum compound test method based on visual network, wherein comprising steps of
A, inquiry and integrate second order metabolism mass spectrometry database, obtain training sample data collection S；
B, the mass spectral intensities of each sample in the training sample data collection S are normalized, are obtained new
Sample data set T；
C, visualized operation is carried out to each sample in the new sample data set T, constructs visual network, obtains
Visual network data set G；
D, network global characteristics are extracted from each visual network in the visual network data set G, acquisition can
Set of eigenvectors F depending on changing Network data set G；
E, using described eigenvector collection F as the input of SVM, output of the corresponding compound name as SVM, to described
New sample data set T is trained, and obtains detection model P, for detecting to second order metabolism mass spectrum compound.
Preferably, the second order based on visual network is metabolized mass spectrum compound test method, wherein in step A
In, the training sample data collection S={ S1,S2,…,SN, any sample SN=[(m1,i1),(m2,i2),…(md,
id)], the mdAnd idThe karyoplasmic ratio numerical value and strength values of respectively the d articles spectral line.
Preferably, the second order based on visual network is metabolized mass spectrum compound test method, wherein the step
B is specifically included:
B1, the mass spectral intensities of each sample in the training sample data collection S are normalized, make each sample
This mass spectral intensities normalize to 0~1000；
B2, all spectral lines of each sample are detected, when detecting the intensity of spectral line lower than 50, then described in deletion
Spectral line, to obtain new sample data set T={ T1,T2,…,TN}。
Preferably, the second order based on visual network is metabolized mass spectrum compound test method, wherein the step
C is specifically included:
C1, every spectral line in each sample in new sample data set T is considered as the node in visual network；
C2, the empty matrix H [] for creating a n*n are used to record the connection of each node, if two nodes are connected
It is then denoted as 1, is denoted as 0 if not connected；
C3, whole adjacent nodes are connected, simultaneously for all two non-adjacent node i (mi,ii) and j (mj,ij), when
Node k (m between described two nodesk,ik), wherein i < k < j meets:
Then connect described two nonneighbor node i (mi,ii) and j (mj,ij)；
C4, visual network is constructed according to the node connection relationship, obtains visual network data set G={ G1,
G2,......GN}。
Preferably, the second order based on visual network is metabolized mass spectrum compound test method, wherein in step D
In, the network global characteristics specifically include:
Node number, network average degree, the cluster coefficients of network, the diameter of network, network average shortest path length and
The density of network.
Preferably, the second order based on visual network is metabolized mass spectrum compound test method, wherein in step D
In, described eigenvector collection F={ F1,F2,.......FN, the feature vector F of any visual networkN=N, DE, C,
DIA, MP, DEN }, wherein the N is node number, DE is network average degree, C is the cluster coefficients of network, DIA is network
Diameter, MP are the shortest path of network, the density that DEN is network.
Preferably, the second order based on visual network is metabolized mass spectrum compound test method, wherein the step
After A further include:
A1, building Decoy data set S corresponding with training sample data collection SD。
Preferably, the second order based on visual network is metabolized mass spectrum compound test method, wherein the step
After E further include:
F, by the Decoy data set SDAs test set, cross validation is carried out to the detection model P.
Preferably, the second order based on visual network is metabolized mass spectrum compound test method, wherein the step
After E further include:
G, using SVM kernel function as compound similarity evaluation function, and evaluation result is normalized to 0~1.
Preferably, the second order based on visual network is metabolized mass spectrum compound test method, wherein the step
G is specifically included:
G1, successively use linear kernel function, radial base core and Sigmoid kernel function pre- to compound progress similarity
It surveys；
G2, the most accurate function of prediction result is chosen as compound similarity evaluation function, and by evaluation result normalizing
Change to 0~1.
The utility model has the advantages that the present invention by existing second order be metabolized mass spectrometry database based on, by each compound second order
Mass spectrometric data carries out building visual network operation, and the input that its network global characteristics is extracted as its respective compound is special
Sign, and it is trained by SVM, compound test model is obtained, and by building Decoy test set to detection model
Cross validation is carried out, guarantees the reliability and accuracy of detection model, it is made to can be used for practical second order metabolism mass spectrum compound
Detection.Second order metabolism mass spectrum compound test method provided by the invention is realized simple and significantly improves the inspection of compound
Degree of testing the speed and precision.
Detailed description of the invention
Fig. 1 is that a kind of second order based on visual network of the present invention is metabolized the method preferred embodiment of mass spectrum compound test
Flow chart.
Fig. 2 is the first mass spectrographic data mode schematic diagram of metabolin.
Fig. 3 is mass spectrographic second of data mode schematic diagram of metabolin.
Specific embodiment
The present invention provides a kind of second order metabolism mass spectrum compound test method based on visual network, of the invention to make
Purpose, technical solution and effect are clearer, clear, referring to the drawings and give an actual example that the present invention is described in more detail.
It should be appreciated that the specific embodiments described herein are merely illustrative of the present invention, it is not intended to limit the present invention.
Referring to Fig. 1, Fig. 1 be a kind of second order based on visual network of the present invention be metabolized mass spectrum compound test method compared with
The flow chart of good embodiment, as shown, itself comprising steps of
S100, inquiry simultaneously integrate second order metabolism mass spectrometry database, obtain training sample data collection S；
Specifically, mass spectral analysis (Mass Spectrometry, MS) be the mostly important research tool of metabolism group it
One, second order mass spectrum therein can effectively identify different metabolites, and accurately measure it with respect to solubility.Base provided by the invention
It is metabolized mass spectrum compound test method in the second order of visual network, is metabolized mass spectrometric data by inquiring existing second order first
Then library, such as MassBank, HMBD integrate the second order metabolism mass spectrometric data and obtain training sample data collection S.
Further, the training sample data collection S={ S1,S2,…,SN, any sample SN=[(m1,i1),(m2,
i2),…(md,id)], the mdAnd idThe karyoplasmic ratio numerical value and strength values of respectively the d articles spectral line.Preferably, such as Fig. 2 and figure
Shown in 3, Fig. 2 and Fig. 3 are respectively the mass spectrographic two kinds of different data forms of metabolin.
Further, after the step S100 further include:
S101, building Decoy data set S corresponding with training sample data collection SD。
Specifically, the present invention can pass through Passatutto software building Decoy corresponding with training sample data collection S
Data set SD。
Further, the step S200, the mass spectral intensities of each sample in the training sample data collection S are returned
One change processing, obtains new sample data set T, specifically includes:
S210, the mass spectral intensities of each sample in the training sample data collection S are normalized, are made each
The mass spectral intensities of sample normalize to 0~1000；
S220, all spectral lines of each sample are detected, when detecting the intensity of spectral line lower than 50, then deletes institute
Spectral line is stated, to obtain new sample data set T={ T1,T2,…,TN}。
Specifically, through the above steps to each metabolism mass spectrum sample S in training sample data collection SNIt has carried out pre-
Processing, deletes SNIntensity is lower than 50 spectral line, obtains new sample data set T={ T1,T2,…,TN}.By to each generation
Thank to mass spectrum sample SNIt is pre-processed, more accurate detection model can be constructed, to effectively improve the detection essence of compound
Degree.
Further, in the present invention, the step S300, in the new sample data set T each sample carry out
Visualized operation constructs visual network, obtains visual network data set G, specifically include:
S310, every spectral line in each sample in new sample data set T is considered as the node in visual network；
Specifically, to each of new sample data set T sample TN ,=[(m1,i1),(m2,i2),…(mn,in)],
Visualized operation is carried out, visual network is constructed, each of them spectral line is considered as the node in visual network；
S320, the empty matrix H [] for creating a n*n are used to record the connection of each node, if two nodes are connected
It connects, is denoted as 1, be denoted as 0 if not connected；For example, H [1,2]=0, this indicates that node 1 and node 2 do not connect.
S330, whole adjacent nodes, i.e. H [n, n+1]=1 are connected；Simultaneously for all two non-adjacent node i (mi,
ii) and j (mj,ij), as the node k (m between described two nodesk,ik), wherein i < k < j meets:
When, illustrate that node i can see node j, then connects described two nonneighbor nodes
i(mi,ii) and j (mj,ij)；
C4, visual network is constructed according to the node connection relationship, obtains visual network data set G={ G1,
G2,......GN}；
Specifically, using NetworkX software building network, visual network G is obtainedN: G.add_edges_from
(H}
nx.draw(G)
By the visual network GNConstitute visual network data set G={ G1,G2,......GN}。
S400, network global characteristics are extracted from each visual network in the visual network data set G, obtain
The set of eigenvectors F of visual network data set G；
Specifically, the network global characteristics specifically include: node number, network average degree, the cluster coefficients of network, net
The density of the diameter of network, the average shortest path length of network and network.
Further, described eigenvector collection F={ F1,F2,.......FN, the feature of any visual network to
Measure FN={ N, DE, C, DIA, MP, DEN }, wherein the N is node number；DE is network average degree；C=nx.average_
Clustering (G) is the cluster coefficients of network；DIA=nx.diameter (G) is the diameter of network；MP=
Nx.average_shortest_path_length (G) is the shortest path of network；DEN=nx.density (G) is network
Density.
S500, using described eigenvector collection F as the input of SVM, output of the corresponding compound name as SVM is right
The new sample data set T is trained, and obtains detection model P, for detecting to second order metabolism mass spectrum compound.
Specifically, by set of eigenvectors F={ F1,F2,.......FNInput as SVM, corresponding compound name
As output, test set data are trained, obtain prediction model P:
Clf=svm. ()
clf.fit(FN,NAME)。
Further, in the present invention, after the step S500 further include:
S600, by the Decoy data set SDAs test set, cross validation is carried out to the detection model P.
Preferably, the present invention by Decoy database carry out that cross validation has been effectively ensured detection method can
By property and stability.
Further, in the present invention, after the step S500 further include:
S700, using SVM kernel function as compound similarity evaluation function, and evaluation result is normalized to 0~1.
Specifically, successively compound is carried out using linear kernel function, radial base core and Sigmoid kernel function similar
Degree prediction；The most accurate function of prediction result is chosen as compound similarity evaluation function, and evaluation result is normalized to 0
~1.
In conclusion the present invention by existing second order be metabolized mass spectrometry database based on, by each compound second order
Mass spectrometric data carries out building visual network operation, and the input that its network global characteristics is extracted as its respective compound is special
Sign, and it is trained by SVM, compound test model is obtained, and by building Decoy test set to detection model
Cross validation is carried out, guarantees the reliability and accuracy of detection model, it is made to can be used for practical second order metabolism mass spectrum compound
Detection.Compared with prior art, the second order metabolism mass spectrum compound test method tool provided by the invention based on visual network
It has the advantage that
The first, input data uses more advanced second order ms, the precision of prediction of compound can be effectively improved, in data
While amount increases, the present invention relies only on spectral line without complicated mass spectrum pretreatment and peak detection, the building of visual network
Basic information karyoplasmic ratio and intensity, realize simple, effectively expanded the application range of this method, reduce processing difficulty at
This, has been obviously improved detection speed and precision；
The second, the global characteristics that the present invention chooses visual network are used as input, pass through SVM and carry out the excellent of detection model
Change, solves the disadvantage that other net mate algorithms need a large amount of manual interventions, while being handed over by Decoy database
Fork verifying ensure that the reliability and stability of algorithm；
Third, the present invention respective optimum prediction model can be trained according to different data, and provide two compounds it
Between similarity, reflect the correlation degree between different compounds, can be used for studying its potential Biochemical Mechanism.In addition, our
Method can also be used in the analysis of the data such as genomics, proteomics.
It should be understood that the application of the present invention is not limited to the above for those of ordinary skills can
With improvement or transformation based on the above description, wanted for example, all these modifications and variations all should belong to right appended by the present invention
The protection scope asked.
Claims (9)
1. a kind of second order based on visual network is metabolized mass spectrum compound test method, which is characterized in that comprising steps of
A, inquiry and integrate second order metabolism mass spectrometry database, obtain training sample data collection S；
B, the mass spectral intensities of each sample in the training sample data collection S are normalized, obtain new sample
Data set T；
C, visualized operation is carried out to each sample in the new sample data set T, constructs visual network, obtained visual
Change Network data set G；
D, network global characteristics are extracted from each visual network in the visual network data set G, are visualized
The set of eigenvectors F of Network data set G；
E, using described eigenvector collection F as the input of SVM, output of the corresponding compound name as SVM, to described new
Sample data set T is trained, and obtains detection model P, for detecting to second order metabolism mass spectrum compound；
The step B is specifically included:
B1, the mass spectral intensities of each sample in the training sample data collection S are normalized, make each sample
Mass spectral intensities normalize to 0~1000；
B2, all spectral lines of each sample are detected, when detecting the intensity of spectral line lower than 50, then deletes the spectrum
Line, to obtain new sample data set T={ T1,T2,…,TN}。
2. the second order according to claim 1 based on visual network is metabolized mass spectrum compound test method, feature exists
In, in step, the training sample data collection S={ S1,S2,…,SN, any sample SN=[(m1,i1),(m2,
i2),…(md,id)], the mdAnd idThe karyoplasmic ratio numerical value and strength values of respectively the d articles spectral line.
3. the second order according to claim 1 based on visual network is metabolized mass spectrum compound test method, feature exists
In the step C is specifically included:
C1, every spectral line in each sample in new sample data set T is considered as the node in visual network；
C2, the empty matrix H [] for creating a n*n remember if two nodes are connected for recording the connection of each node
It is 1, is denoted as 0 if not connected, wherein n is the order of empty matrix H []；
C3, whole adjacent nodes are connected, simultaneously for all two non-adjacent node i (mi,ii) and j (mj,ij), when in institute
State the node k (m between two nodesk,ik), wherein i < k < j meets:
Then connect described two nonneighbor node i (mi,ii) and j (mj,ij), wherein mi
And iiThe karyoplasmic ratio numerical value and strength values of respectively i-th spectral line, mjAnd ijRespectively the karyoplasmic ratio numerical value of j-th strip spectral line with
Strength values, mkAnd ikThe respectively karyoplasmic ratio numerical value and strength values of kth spectral line；
C4, visual network is constructed according to the node connection relationship, obtains visual network data set G={ G1,G2,
......GN}。
4. the second order according to claim 1 based on visual network is metabolized mass spectrum compound test method, feature exists
In in step D, the network global characteristics specifically include:
Node number, network average degree, the cluster coefficients of network, the diameter of network, network average shortest path length and network
Density.
5. the second order according to claim 1 based on visual network is metabolized mass spectrum compound test method, feature exists
In, in step D, described eigenvector collection F={ F1,F2,.......FN, the feature vector F of any visual networkN
={ N, DE, C, DIA, MP, DEN }, wherein the cluster coefficients that the N is node number, DE is network average degree, C is network,
DIA is the diameter of network, MP is the shortest path of network, the density that DEN is network.
6. the second order according to claim 1 based on visual network is metabolized mass spectrum compound test method, feature exists
In after the step A further include:
A1, building Decoy data set S corresponding with training sample data collection SD。
7. the second order according to claim 6 based on visual network is metabolized mass spectrum compound test method, feature exists
In after the step E further include:
F, by the Decoy data set SDAs test set, cross validation is carried out to the detection model P.
8. the second order according to claim 1 based on visual network is metabolized mass spectrum compound test method, feature exists
In after the step E further include:
G, using SVM kernel function as compound similarity evaluation function, and evaluation result is normalized to 0~1.
9. the second order according to claim 8 based on visual network is metabolized mass spectrum compound test method, feature exists
In the step G is specifically included:
G1, linear kernel function, radial base core and Sigmoid kernel function is successively used to carry out similarity prediction to compound；
G2, the most accurate function of prediction result is chosen as compound similarity evaluation function, and evaluation result is normalized to 0
~1.
Priority Applications (2)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201610925871.3A CN106528668B (en) | 2016-10-23 | 2016-10-23 | A kind of second order metabolism mass spectrum compound test method based on visual network |
| PCT/CN2016/111336 WO2018072306A1 (en) | 2016-10-23 | 2016-12-21 | Visualization network-based two-stage metabolite mass spectrometry detection method for compound |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201610925871.3A CN106528668B (en) | 2016-10-23 | 2016-10-23 | A kind of second order metabolism mass spectrum compound test method based on visual network |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN106528668A CN106528668A (en) | 2017-03-22 |
| CN106528668B true CN106528668B (en) | 2018-12-25 |
Family
ID=58291637
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201610925871.3A Expired - Fee Related CN106528668B (en) | 2016-10-23 | 2016-10-23 | A kind of second order metabolism mass spectrum compound test method based on visual network |
Country Status (2)
| Country | Link |
|---|---|
| CN (1) | CN106528668B (en) |
| WO (1) | WO2018072306A1 (en) |
Families Citing this family (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108846254B (en) * | 2018-06-27 | 2021-08-24 | 哈尔滨工业大学（深圳） | Second-order metabolic mass spectrometry multi-compound detection method, storage medium and server |
| CN114594171B (en) * | 2020-12-03 | 2023-12-15 | 中国科学院大连化学物理研究所 | Metabolome deep annotation method |
| CN114624340B (en) * | 2020-12-08 | 2022-11-08 | 中国科学院大连化学物理研究所 | Identification method of plant disease and insect pest mediated plant resistance related secondary metabolites in plant |
Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| EP2083265A1 (en) * | 2006-10-31 | 2009-07-29 | Keio University | Method of classifying protein/compound pairs |
| CN103235030A (en) * | 2013-03-25 | 2013-08-07 | 江苏易谱恒科技有限公司 | Distillate spirit brand identification method based on support vector machine and time-of-flight mass spectrometry |
| CN103703371A (en) * | 2011-04-29 | 2014-04-02 | 癌症预防和治疗有限公司 | Methods of identification and diagnosis of lung diseases using classification systems and kits thereof |
| CN105718744A (en) * | 2016-01-25 | 2016-06-29 | 深圳大学 | Metabolic mass spectrum screening method and system based on the deep learning |
| CN105893790A (en) * | 2016-06-23 | 2016-08-24 | 西南大学 | Classification method for mass spectrum deficiency protein data |
Family Cites Families (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| BRPI0506117A (en) * | 2005-10-14 | 2007-07-03 | Fundacao Oswaldo Cruz | diagnostic method based on proteinaceous and / or genetic patterns by support vectors applied to mass spectrometry |
| CN104376325A (en) * | 2014-10-30 | 2015-02-25 | 中国科学院半导体研究所 | Method for building near-infrared qualitative analysis model |
| CN105138866A (en) * | 2015-08-12 | 2015-12-09 | 广东顺德中山大学卡内基梅隆大学国际联合研究院 | Method for identifying protein functions based on protein-protein interaction network and network topological structure features |
-
2016
- 2016-10-23 CN CN201610925871.3A patent/CN106528668B/en not_active Expired - Fee Related
- 2016-12-21 WO PCT/CN2016/111336 patent/WO2018072306A1/en active Application Filing
Patent Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| EP2083265A1 (en) * | 2006-10-31 | 2009-07-29 | Keio University | Method of classifying protein/compound pairs |
| CN103703371A (en) * | 2011-04-29 | 2014-04-02 | 癌症预防和治疗有限公司 | Methods of identification and diagnosis of lung diseases using classification systems and kits thereof |
| CN103235030A (en) * | 2013-03-25 | 2013-08-07 | 江苏易谱恒科技有限公司 | Distillate spirit brand identification method based on support vector machine and time-of-flight mass spectrometry |
| CN105718744A (en) * | 2016-01-25 | 2016-06-29 | 深圳大学 | Metabolic mass spectrum screening method and system based on the deep learning |
| CN105893790A (en) * | 2016-06-23 | 2016-08-24 | 西南大学 | Classification method for mass spectrum deficiency protein data |
Also Published As
| Publication number | Publication date |
|---|---|
| WO2018072306A1 (en) | 2018-04-26 |
| CN106528668A (en) | 2017-03-22 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN105718744B (en) | A kind of metabolism mass spectrum screening method and system based on deep learning | |
| Araya‐Salas et al. | warbleR: an R package to streamline analysis of animal acoustic signals | |
| McFee et al. | Analyzing Song Structure with Spectral Clustering. | |
| CN109299380B (en) | Exercise personalized recommendation method based on multi-dimensional features in online education platform | |
| CN110135459B (en) | Zero sample classification method based on double-triple depth measurement learning network | |
| CN111222847B (en) | Open source community developer recommendation method based on deep learning and unsupervised clustering | |
| CN110362660A (en) | A kind of Quality of electronic products automatic testing method of knowledge based map | |
| CN103049526B (en) | Based on the cross-media retrieval method of double space study | |
| CN105389326B (en) | Image labeling method based on weak matching probability typical relevancy models | |
| CN108256590B (en) | A kind of similar traveler recognition methods based on compound first path | |
| CN106528668B (en) | A kind of second order metabolism mass spectrum compound test method based on visual network | |
| CN110309268A (en) | A kind of cross-language information retrieval method based on concept map | |
| CN102663447B (en) | Cross-media searching method based on discrimination correlation analysis | |
| CN111161249B (en) | Unsupervised medical image segmentation method based on domain adaptation | |
| CN106203483A (en) | A kind of zero sample image sorting technique of multi-modal mapping method of being correlated with based on semanteme | |
| CN109255289A (en) | A kind of across aging face identification method generating model based on unified formula | |
| Ollier et al. | Orthonormal transform to decompose the variance of a life-history trait across a phylogenetic tree | |
| CN110348494A (en) | A kind of human motion recognition method based on binary channels residual error neural network | |
| CN103164537B (en) | A kind of method of search engine logs data mining of user oriented information requirement | |
| CN113705570A (en) | Few-sample target detection method based on deep learning | |
| Frey et al. | Most powerful rank tests for perfect rankings | |
| CN113408616B (en) | Spectral classification method based on PCA-UVE-ELM | |
| CN112350745B (en) | Sorting method of frequency hopping communication radio station | |
| Seo et al. | Computing Interdisciplinarity of Scholarly Objects using an Author-Citation-Text Model. | |
| CN114970684A (en) | Community detection method for extracting network core structure by combining VAE |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant | ||
| CF01 | Termination of patent right due to non-payment of annual fee |
Granted publication date: 20181225 Termination date: 20191023 |
|
| CF01 | Termination of patent right due to non-payment of annual fee |