http://rdf.ncbi.nlm.nih.gov/pubchem/patent/CN-107622182-B
Outgoing Links
| Predicate | Object |
|---|---|
| classificationIPCInventive | http://rdf.ncbi.nlm.nih.gov/pubchem/patentipc/G06N3-08 http://rdf.ncbi.nlm.nih.gov/pubchem/patentipc/G16B15-00 http://rdf.ncbi.nlm.nih.gov/pubchem/patentipc/G16B40-00 |
| filingDate | 2017-08-04-04:00^^<http://www.w3.org/2001/XMLSchema#date> |
| grantDate | 2020-10-09-04:00^^<http://www.w3.org/2001/XMLSchema#date> |
| publicationDate | 2020-10-09-04:00^^<http://www.w3.org/2001/XMLSchema#date> |
| publicationNumber | CN-107622182-B |
| titleOfInvention | Method and system for predicting local structural features of protein |
| abstract | The invention relates to the field of bioinformatics, and discloses a method and a system for predicting local structural features of a protein, which are used for improving the prediction accuracy by utilizing a deep learning technology, providing key reference information for the prediction of a tertiary structure of the protein, and solving the problems of high cost and low efficiency caused by the determination of the tertiary structure of the protein by a biological experiment method. The method uniformly constructs a characteristic sequence of each protein sequence in a sample set to be used as input of a training model, the training model adopts a deep neural network model with 3 hidden layers based on stack-type sparse self-coding, a dropout method is applied to the hidden layers of the whole network, and certain neurons in the hidden layers do not work randomly to reduce overfitting of the model; and optimizing the weight parameters of the training model through the training set to ensure that the value of the constructed loss function reaches the minimum, and then correspondingly predicting the solvent accessibility or residue contact number of each residue in the protein sequence according to the trained network model. |
| priorityDate | 2017-08-04-04:00^^<http://www.w3.org/2001/XMLSchema#date> |
| type | http://data.epo.org/linked-data/def/patent/Publication |
Incoming Links
Total number of triples: 30.