CN107273714A - The ATP binding site estimation methods of conjugated protein sequence and structural information - Google Patents
The ATP binding site estimation methods of conjugated protein sequence and structural information Download PDFInfo
- Publication number
- CN107273714A CN107273714A CN201710424110.4A CN201710424110A CN107273714A CN 107273714 A CN107273714 A CN 107273714A CN 201710424110 A CN201710424110 A CN 201710424110A CN 107273714 A CN107273714 A CN 107273714A
- Authority
- CN
- China
- Prior art keywords
- atp
- information
- protein
- amino acid
- winsize
- 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
Links
Classifications
-
- G—PHYSICS
- G16—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR SPECIFIC APPLICATION FIELDS
- G16B—BIOINFORMATICS, i.e. INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR GENETIC OR PROTEIN-RELATED DATA PROCESSING IN COMPUTATIONAL MOLECULAR BIOLOGY
- G16B20/00—ICT specially adapted for functional genomics or proteomics, e.g. genotype-phenotype associations
-
- G—PHYSICS
- G16—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR SPECIFIC APPLICATION FIELDS
- G16B—BIOINFORMATICS, i.e. INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR GENETIC OR PROTEIN-RELATED DATA PROCESSING IN COMPUTATIONAL MOLECULAR BIOLOGY
- G16B5/00—ICT specially adapted for modelling or simulations in systems biology, e.g. gene-regulatory networks, protein interaction networks or metabolic networks
Landscapes
- Physics & Mathematics (AREA)
- Health & Medical Sciences (AREA)
- Life Sciences & Earth Sciences (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- General Health & Medical Sciences (AREA)
- Biophysics (AREA)
- Spectroscopy & Molecular Physics (AREA)
- Molecular Biology (AREA)
- Medical Informatics (AREA)
- Bioinformatics & Computational Biology (AREA)
- Biotechnology (AREA)
- Evolutionary Biology (AREA)
- Analytical Chemistry (AREA)
- Proteomics, Peptides & Aminoacids (AREA)
- Genetics & Genomics (AREA)
- Chemical & Material Sciences (AREA)
- Physiology (AREA)
- Peptides Or Proteins (AREA)
Abstract
Site estimation method is bound the invention discloses the ATP of a kind of conjugated protein sequence and structural information, the sequence and structural information of protein are read first, the characteristic vector of each amino acid residue is extracted from protein sequence and structural information；Multiple down-sampling is carried out to non ATP bindings bit point sample using random down-sampling technology, the non ATP bindings bit point sample set that each down-sampling is obtained trains a SVM submodel after merging with ATP bindings bits point sample set, obtains multiple SVM submodels；Above-mentioned multiple SVM submodels are carried out to integrated, the final forecast model of formation using average integrated approach；Carry out judging the amino acid residue whether as ATP bindings site by the way that the characteristic vector of any amino acid residue to be predicted is input in final forecast model；The ATP for finally showing prediction in three dimensions using PyMOL softwares binds site.This method not only significantly reduces the scale of training set, and improves the interpretation and precision of prediction of forecast model.
Description
Technical field
The present invention relates to Bioinformatics Prediction protein-ATP interactions field, specifically a kind of conjugated protein sequence
Row and the ATP binding site estimation methods of structural information.
Background technology
Reciprocation between protein and ATP is of common occurrence in vital movement, and it is widely present in substantial amounts of
In life entity.This reciprocation is usually expressed as the fixed residue of some of ATP binding protein (i.e. ATP binds site),
ATP is cooperated jointly with protein and provides required energy for vital movement.ATP bindings site in protein is usual
It is also the important target spot of chemicals.Want to be fully clarified yet that the process of vital movement, especially about protein and ATP life
The details of activity, the ATP bindings site being precisely located in protein has important biomolecule meaning.
However, determining that the ATP bindings site in protein needs to take a substantial amount of time by the method for Bioexperiment
And fund, and it is less efficient；Moreover, with the continuous propulsion developed rapidly with technique for crystallization of proteins of sequencing technologies, egg
A large amount of protein informations for not carrying out ATP bindings site demarcation are had accumulated which in white matter group.Therefore bioinformatics is applied
Relevant knowledge, directly from protein sequence and structure, research and develop a kind of ATP bindings bits that can be in Accurate Prediction protein
The intelligence computation method of point has active demand, and is given birth to for finding and recognizing protein structure, physiological function and relevant ATP
The details of life activity suffers from great Research Significance.
At present, also it is short of very much for the high precision forecasting models for binding site of the ATP based on protein information.By consulting
Pertinent literature, it is found that the computation model for being designed specifically to predict ATP bindings site in protein at present is all from protein
What sequence information set out, they include：ATPint(Chauhan,Jagat S., Nitish K.Mishra,and Gajendra
PS Raghava."Identification of ATP binding residues of a protein from its
primary sequence."BMC bioinformatics 10.1(2009):434.)、 ATPsite(Chen,Ke,Marcin
J.Mizianty,and Lukasz Kurgan."ATPsite: sequence-based prediction of ATP-
binding residues."Proteome science 9.1(2011): S4.)、targetATP(Yu,Dong-Jun,et
al."Improving protein-ATP binding residues prediction by boosting SVMs with
random under-sampling."Neurocomputing 104 (2013):180-190.), and TargetATPsite
(Yu,Dong‐Jun,et al."TargetATPsite:A template‐free method for ATP‐binding
sites prediction with residue evolution image sparse representation and
classifier ensemble."Journal of computational chemistry 34.11(2013):974-985.)
Deng.They by using different protein sequence characteristics information (such as：Protein evolution information, secondary protein structure letter
Breath, protein solvent accessibility information etc.) combination supporting vector machine (SVM) algorithm trains prediction on different data acquisition systems
Model, achieves certain precision of prediction.
Although having been achieved with part achievement in the ATP binding site estimation researchs based on protein information, the Xiang Yan
Study carefully task and be also far from end.Existing achievement in research is all from protein sequence information, not using albumen
The structural information of matter.In fact, the information content included in protein structure is far longer than in sequence.Conjugated protein sequence with
Structural information is to improve the effective way that ATP binds site estimation precision.Tied up in addition, protein sequence information can not provide ATP
The displaying directly perceived that anchor point predicts the outcome, so as to cause the existing ATP binding site estimation models based on protein sequence
Interpretation is poor.Therefore ATP binding site estimation tasks are in the urgent need to further research.
The content of the invention
Site estimation side is bound it is an object of the invention to provide the ATP of a kind of conjugated protein sequence and structural information
Method, for solving the problems, such as in the above-mentioned ATP binding site estimations based on protein information due to unused protein matter structural information
Caused by precision of prediction apart from the shortcoming that practical application gap is larger and interpretation is poor.
The technical solution for realizing the object of the invention is：A kind of ATP bindings of conjugated protein sequence and structural information
Site estimation method, comprises the following steps：
Step 1：Feature extraction, according to protein sequence to be predicted and structural information, using PSI-BLAST, PSIPRED,
SANN, SSITE and TMSITE program extract respectively the protein evolution information, secondary structure information, solvent accessibility information,
Sequence template match information and stay in place form match information, build each amino acid using sliding window technique on this basis
The characteristic vector of residue；
Step 2：Using random down-sampling technology, multiple down-sampling is carried out to non ATP bindings bit point sample, obtains multiple non-
ATP bindings bits point sample set, after each non ATP bindings bit point sample set is merged with ATP bindings bits point sample set
A SVM submodel is trained, multiple SVM submodels are obtained；
Step 3：It is using average integrated approach, the multiple SVM submodels obtained in step 2 progress is integrated, obtain final
ATP binding site estimation model, for predict the ATP in protein to be predicted bind site；And
Step 4：The ATP for showing prediction in three dimensions using PyMol softwares binds site.
Further, in above-mentioned steps 1, for the protein that any one is made up of n amino acid residue, by using
PSI-BLAST Program extractions obtain the evolution information characteristics of the protein, i.e. position-specific scoring matrices, and size is n × 20；
Sigmoid functions, i.e. f (x)=1/ (1+e are used to the position-specific scoring matrices again-x), standardized line by line, so
The Evolution matrix of each amino acid residue is obtained for winsize sliding window using size afterwards, by Evolution matrix
Pull into the characteristic vector that length is 20 × winsize.
Further, in above-mentioned step 1, the protein being made up of n amino acid is input to PSIPRED journeys
Sequence, obtains Secondary structure feature, that is, the secondary structure probability matrix predicted, matrix size is n × 3；Reuse length
The sliding window for winsize is spent, the secondary structure information matrix of each amino acid residue is obtained；Finally by secondary structure information
Matrix pulls into the characteristic vector that length is 3 × winsize.
Further, in above-mentioned step 1, the protein being made up of n amino acid is input to SANN programs,
The solvent accessibility feature of protein sequence is obtained, that is, the solvent accessibility probability matrix predicted, matrix size is n × 3；Make again
The sliding window for being winsize with length, obtains the solvent accessibility information matrix of each amino acid residue；Finally can by solvent
And property information matrix pulls into the characteristic vector that length is 3 × winsize.
Further, in above-mentioned step 1, the protein being made up of n amino acid is input to SSITE programs,
Obtain a length and include protein sequence template matches information vector for n；Reuse the sliding window that length is winsize
Mouthful, the characteristic vector of the sequence template match information of each amino acid residue is obtained, the length of characteristic vector is winsize.
Further, in above-mentioned step 1, the protein being made up of n amino acid is input to TMSITE journeys
Sequence, obtains a length and includes protein structure template matches information vector for n；Reuse the slip that length is winsize
Window, obtains the characteristic vector of the stay in place form match information of each amino acid residue, the length of characteristic vector is winsize.
The present invention compared with prior art, its remarkable advantage：1. improve the precision of prediction of model：From protein sequence and knot
Structure information is set out, and extraction has obtained multiple visual angle characteristics, excavates more effective information of forecastings, uses random down-sampling technology
It is combined with average integrated technology, it is to avoid the disequilibrium between non ATP binding site and ATP bindings bits point samples is brought
Interference, improve prediction protein in ATP bind site precision of prediction；
2. the interpretation of lift scheme：Intuitively shown in three dimensions by using PyMOL softwares high-precision
ATP binds site estimation result, improves the interpretation of model.
Brief description of the drawings
Accompanying drawing 1 is the ATP bindings site of conjugated protein sequence information feature of the present invention and protein structural information feature
The schematic diagram of Forecasting Methodology.
Embodiment
Accompanying drawing gives the Forecasting Methodology system structure diagram of the present invention.With reference to shown in accompanying drawing, according to the reality of the present invention
Example is applied, the ATP binding site estimation methods of a kind of conjugated protein sequence and structural information include following steps：
First, the protein is extracted respectively using PSI-BLAST, PSIPRED, SANN, SSITE and TMSITE program to enter
Change information, secondary structure information, solvent accessibility information, sequence template match information and stay in place form match information；Secondly,
Using sliding window technique, from protein evolution information, secondary structure information, solvent accessibility information, sequence template matching letter
The various visual angles characteristic vector of each amino acid residue is extracted in breath and stay in place form match information, and they are carried out serially
Combination obtains the characteristic vector eventually for prediction；Again, using random down-sampling technology, non ATP bindings bit point sample is entered
The multiple down-sampling of row, the non ATP bindings bit point sample set that each down-sampling is obtained constitutes one with ATP bindings bits point sample
Training set, a SVM submodel is trained on the training set, multiple SVM submodels are obtained；Then, using average integrated approach,
Above-mentioned multiple SVM submodels progress are integrated, obtain final ATP binding site estimation models；Finally obtaining ATP bindings bits
After point prediction result, three-dimensional structure displaying is carried out using PyMOL softwares.
Below in conjunction with shown in accompanying drawing, aforementioned process is more specifically described.
Step 1：Feature extraction
For the protein for including n amino acid residue, protein evolution can obtain by PSI-BLAST programs and believe
Cease feature, i.e. position-specific scoring matrices (Position Specific Scoring Matrix, PSSM), size be n ×
20 (n rows 20 are arranged), expression matrix form is changed into by protein sequence information, as follows：
Each value in PSSM is normalized：
Using the sliding window that size is winsize, to extract the PSSM eigenmatrixes of each amino acid residue：
Then, the eigenmatrix of the amino acid residue is pulled into the characteristic vector that dimension is 20 × winsize：
For the protein for including n amino acid residue, its secondary structure can be obtained by PSIPRED programs and believed
Cease feature, i.e. prediction probability matrix (Predicted Secondary Structure, PSS), size is n × 3 (n rows 3 are arranged)：
Using above-mentioned an equal amount of sliding window technique, the PSS eigenmatrixes of each amino acid residue can be obtained：
Then, the PSS eigenmatrixes of the amino acid residue are pulled into the characteristic vector that dimension is 3 × winsize：
For the protein for including n amino acid residue, its solvent accessibility feature can obtain by SANN programs,
That is prediction probability matrix (Predicted Solvent Accessibility, PSA), size is n × 3 (n rows 3 are arranged)：
Using above-mentioned an equal amount of sliding window technique, the PSA eigenmatrixes of each amino acid residue can be obtained：
Then, the PSA eigenmatrixes of the amino acid residue are pulled into the characteristic vector that dimension is 3 × winsize：
For the protein for including n amino acid residue, its sequence template matching letter can obtain by SSITE programs
Feature, i.e. sequence template voting results (Sequence Template Voting Results, SeqTVR)：
SeqTVR=(seqtvr1,…,seqtvri,…,seqtvrn) (11)
Using above-mentioned an equal amount of sliding window technique, can obtain the SeqTVR features of each amino acid residue to
Amount：
For the protein for including n amino acid residue, its stay in place form can obtain by TMSITE programs and match
Believe feature, i.e. stay in place form voting results (Structure Template Voting Results, StrTVR)：
StrTVR=(strtvr1,…,strtvri,…,strtvrn) (13)
Using above-mentioned an equal amount of sliding window technique, can obtain the StrTVR features of each amino acid residue to
Amount：
Step 2：Using random down-sampling technology, the sample that site is bound to non ATP carries out multiple down-sampling, will each time
The non ATP bindings bit point subset that down-sampling is obtained constitutes a training set with ATP bindings bits point sample, and instruction is closed in the training set
Practice a SVM submodel, finally give multiple SVM submodels.By to ATP unbundlings site sample set it is multiple it is random under
Sampling, can be under the premise of ensureing that sample distribution is uniform, loss non ATP binding site information as few as possible.
Step 3：The multiple SVM submodels obtained in step 2 are unified into final prediction mould using average Integrated Algorithm
Type, for predicting whether each amino acid residue is ATP bindings site in protein to be predicted.
Step 4：Show that final ATP binds site estimation result in three dimensions using PyMOL softwares.
In summary, it is of the invention compared with existing Forecasting Methodology, this method can effectively using protein sequence with
Two kinds of different visual angles information of structure, and use SVMs (SVM) algorithm effectively utilizes the information of two kinds of different visual angles, and
And this method also possesses the energy for the disequilibrium for solving the problems, such as to be present between non ATP binding site and ATP bindings bits point sample
Power and the ability for intuitively showing ATP binding site estimation results in three dimensions.It is explainable same in enhancing forecast model
When, also final precision of prediction is got a promotion.
Claims (6)
1. a kind of ATP binding site estimation methods of conjugated protein sequence and structural information, it is characterised in that including following step
Suddenly：
Step 1：Feature extraction；According to protein sequence to be predicted and structural information, using PSI-BLAST, PSIPRED, SANN,
SSITE and TMSITE programs extract the protein evolution information, secondary structure information, solvent accessibility information, sequence respectively
Template matches information and stay in place form match information, build each amino acid residue using sliding window technique on this basis
Characteristic vector；
Step 2：Using random down-sampling technology, multiple down-sampling is carried out to non ATP bindings bit point sample, multiple non ATPs are obtained
Bindings bit point sample set, is trained after each non ATP bindings bit point sample set is merged with ATP bindings bits point sample set
One SVM submodel, obtains multiple SVM submodels；
Step 3：It is using average integrated approach, the multiple SVM submodels obtained in step 2 progress is integrated, obtain final ATP
Site estimation model is bound, for predicting that the ATP in protein to be predicted binds site；
Step 4：The ATP for showing prediction in three dimensions using PyMol softwares binds site.
2. the ATP binding site estimation methods of conjugated protein sequence according to claim 1 and structural information, its feature
It is：In the step 1, for the protein that any one is made up of n amino acid residue, by using PSI-BLAST journeys
Sequence extracts the evolution information characteristics for obtaining the protein, i.e. position-specific scoring matrices, and size is n × 20；It is special to position again
Different in nature score matrix uses sigmoid functions, i.e. f (x)=1/ (1+e-x), standardized line by line, x is location specific score
Arbitrary element in matrix, then obtains the Evolution of each amino acid residue using size for winsize sliding window
Matrix, the characteristic vector that length is 20 × winsize is pulled into by Evolution matrix.
3. the ATP binding site estimation methods of conjugated protein sequence according to claim 1 and structural information, its feature
It is：In the step 1, the protein being made up of n amino acid is input to PSIPRED programs, protein is obtained
Second structure characteristic, that is, the secondary structure probability matrix predicted, matrix size is n × 3；Reuse the cunning that length is winsize
Dynamic window, obtains the secondary structure information matrix of each amino acid residue；It is 3 that secondary structure information matrix finally is pulled into length
× winsize characteristic vector.
4. the ATP binding site estimation methods of conjugated protein sequence according to claim 1 and structural information, its feature
It is：In above-mentioned step 1, the protein being made up of n amino acid is input to SANN programs, protein sequence is obtained
The solvent accessibility feature of row, that is, the solvent accessibility probability matrix predicted, matrix size is n × 3；Reusing length is
Winsize sliding window, obtains the solvent accessibility information matrix of each amino acid residue；Finally by solvent accessibility information
Matrix pulls into the characteristic vector that length is 3 × winsize.
5. the ATP binding site estimation methods of conjugated protein sequence according to claim 1 and structural information, its feature
It is：In the step 1, the protein being made up of n amino acid is input to SSITE programs, a length is obtained for n
Include protein sequence template matches information vector；The sliding window that length is winsize is reused, each amino acid is obtained
The characteristic vector of the sequence template match information of residue, the length of characteristic vector is winsize.
6. the ATP binding site estimation methods of conjugated protein sequence according to claim 1 and structural information, its feature
It is：In the step 1, the protein being made up of n amino acid is input to TMSITE programs, obtaining a length is
N's includes protein structure template matches information vector；The sliding window that length is winsize is reused, each amino is obtained
The characteristic vector of the stay in place form match information of sour residue, the length of characteristic vector is winsize.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201710424110.4A CN107273714A (en) | 2017-06-07 | 2017-06-07 | The ATP binding site estimation methods of conjugated protein sequence and structural information |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201710424110.4A CN107273714A (en) | 2017-06-07 | 2017-06-07 | The ATP binding site estimation methods of conjugated protein sequence and structural information |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN107273714A true CN107273714A (en) | 2017-10-20 |
Family
ID=60066491
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201710424110.4A Pending CN107273714A (en) | 2017-06-07 | 2017-06-07 | The ATP binding site estimation methods of conjugated protein sequence and structural information |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN107273714A (en) |
Cited By (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108804867A (en) * | 2018-06-15 | 2018-11-13 | 中国人民解放军军事科学院军事医学研究院 | The model building method of pyrimidine dimer in radiation injury is identified based on Nanopore sequencing technologies |
| CN109147866A (en) * | 2018-06-28 | 2019-01-04 | 南京理工大学 | Residue prediction technique is bound based on sampling and the protein-DNA of integrated study |
| CN109326329A (en) * | 2018-11-14 | 2019-02-12 | 金陵科技学院 | Zinc-binding protein matter action site prediction technique based on integrated study under a kind of unbalanced mode |
| CN110197700A (en) * | 2019-04-16 | 2019-09-03 | 浙江工业大学 | A kind of a-protein TP interconnection method based on differential evolution |
| CN113361752A (en) * | 2021-05-21 | 2021-09-07 | 浙江工业大学 | Protein solvent accessibility prediction method based on multi-view learning |
Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103955628A (en) * | 2014-04-22 | 2014-07-30 | 南京理工大学 | Subspace fusion-based protein-vitamin binding location point predicting method |
| CN104992079A (en) * | 2015-06-29 | 2015-10-21 | 南京理工大学 | Sampling learning based protein-ligand binding site prediction method |
| CN105808975A (en) * | 2016-03-14 | 2016-07-27 | 南京理工大学 | Multi-core-learning and Boosting algorithm based protein-DNA binding site prediction method |
-
2017
- 2017-06-07 CN CN201710424110.4A patent/CN107273714A/en active Pending
Patent Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103955628A (en) * | 2014-04-22 | 2014-07-30 | 南京理工大学 | Subspace fusion-based protein-vitamin binding location point predicting method |
| CN104992079A (en) * | 2015-06-29 | 2015-10-21 | 南京理工大学 | Sampling learning based protein-ligand binding site prediction method |
| CN105808975A (en) * | 2016-03-14 | 2016-07-27 | 南京理工大学 | Multi-core-learning and Boosting algorithm based protein-DNA binding site prediction method |
Cited By (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108804867A (en) * | 2018-06-15 | 2018-11-13 | 中国人民解放军军事科学院军事医学研究院 | The model building method of pyrimidine dimer in radiation injury is identified based on Nanopore sequencing technologies |
| CN109147866A (en) * | 2018-06-28 | 2019-01-04 | 南京理工大学 | Residue prediction technique is bound based on sampling and the protein-DNA of integrated study |
| CN109326329A (en) * | 2018-11-14 | 2019-02-12 | 金陵科技学院 | Zinc-binding protein matter action site prediction technique based on integrated study under a kind of unbalanced mode |
| CN109326329B (en) * | 2018-11-14 | 2020-07-07 | 金陵科技学院 | Zinc binding protein action site prediction method |
| CN110197700A (en) * | 2019-04-16 | 2019-09-03 | 浙江工业大学 | A kind of a-protein TP interconnection method based on differential evolution |
| CN110197700B (en) * | 2019-04-16 | 2021-04-06 | 浙江工业大学 | Protein ATP docking method based on differential evolution |
| CN113361752A (en) * | 2021-05-21 | 2021-09-07 | 浙江工业大学 | Protein solvent accessibility prediction method based on multi-view learning |
| CN113361752B (en) * | 2021-05-21 | 2022-07-26 | 浙江工业大学 | Protein solvent accessibility prediction method based on multi-view learning |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN107273714A (en) | The ATP binding site estimation methods of conjugated protein sequence and structural information | |
| US11514272B2 (en) | Apparatus and method for training classification model and apparatus for performing classification by using classification model | |
| Chang et al. | Identification of the technology life cycle of telematics: A patent-based analytical perspective | |
| CN111476324B (en) | Traffic data labeling method, device, equipment and medium based on artificial intelligence | |
| CN103955628B (en) | The protein vitamin binding site estimation method being merged based on subspace | |
| WO2019242286A1 (en) | Systems and methods for allocating service requests | |
| CN111652279B (en) | Behavior evaluation method and device based on time sequence data and readable storage medium | |
| CN105808975A (en) | Multi-core-learning and Boosting algorithm based protein-DNA binding site prediction method | |
| CN110827236B (en) | Brain tissue layering method, device and computer equipment based on neural network | |
| CN111027600B (en) | Image category prediction method and device | |
| CN109147866A (en) | Residue prediction technique is bound based on sampling and the protein-DNA of integrated study | |
| CN104992079A (en) | Sampling learning based protein-ligand binding site prediction method | |
| CN113569852A (en) | Training method and device of semantic segmentation model, electronic equipment and storage medium | |
| CN114611672A (en) | Model training method, face recognition method and device | |
| EP3913533A2 (en) | Method and apparatus of processing image device and medium | |
| US11881044B2 (en) | Method and apparatus for processing image, device and storage medium | |
| CN112069782B (en) | Document template generation method and device, electronic equipment and storage medium | |
| US20230145129A1 (en) | Generating neural network outputs by enriching latent embeddings using self-attention and cross-attention operations | |
| CN116884475A (en) | Drug target interaction analysis method, device, equipment and storage medium | |
| WO2023011062A1 (en) | Information pushing method and apparatus, device, storage medium, and computer program product | |
| CN115346095A (en) | Visual question answering method, device, equipment and storage medium | |
| CN115146064A (en) | Intention recognition model optimization method, device, equipment and storage medium | |
| CN114861873A (en) | Multi-stage computationally efficient neural network inference | |
| CN113515591A (en) | Text bad information identification method and device, electronic equipment and storage medium | |
| Peng et al. | Identification of incorrect karyotypes using deep learning |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20171020 |
|
| RJ01 | Rejection of invention patent application after publication |