发明名称 |
Method, system and apparatus to predict and/or recognize and/or classify biological sequences |
摘要 |
A method, a system and an apparatus for predicting and/or recognizing and/or classifying biological sequences, specially sequence families with binding site recognition motifs poorly conserved, comprising, advantageously, the use of neural networks rules; providing enhanced and more accurate results; and is preferably used when the biological sequence is a promoter. |
申请公布号 |
US8762119(B2) |
申请公布日期 |
2014.06.24 |
申请号 |
US201012855366 |
申请日期 |
2010.08.12 |
申请人 |
Universidade de Caxias do Sul |
发明人 |
de Avila e Silva Scheila;Laguna Sergio Echeverrigaray;Gerhardt Gunther Johannes Lewezuk |
分类号 |
G06G7/58;G06F7/60;G01N33/48;G01N33/50;G01N31/00;G06F19/24;G06F19/12;G06F19/22 |
主分类号 |
G06G7/58 |
代理机构 |
Smith Risley Tempel Santos LLC |
代理人 |
Colton Laurence P.;Smith Risley Tempel Santos LLC |
主权项 |
1. A method to predict and/or recognize and/or classify biological sequences wherein said biological sequences are predicted, recognized and/or classified by rules extracted from Neural Network (NN) learning process for poorly conserved biological sequences, the process being carried out by a computer and comprising the steps of:
a) conducting NN learning for “X” sequences obtained from a database of sequences of binding site recognition motifs poorly conserved, namely, sequences where the relative position of the motifs are poorly conserved; b) extracting a rule for the “X” sequences of binding site recognition motifs poorly conserved; c) replacing prototype values from NN rule extraction by an integer number for the “X” sequences of binding site recognition motifs poorly conserved; d) analyzing each of the sequences for the “X” sequences of binding site recognition motifs poorly conserved; e) scoring each of the sequences and determining whether each of the sequences is greater than a cut-off value; and f) verifying if each of the sequences is a promoter of “X” sigma factor family, where “X” means the family of a given sequence, “X” being more than 1, and wherein the architecture of the NN learning process comprises between 2 and 5 neurons in a hidden layer of the NN, thereby allowing for the predicting and/or recognizing and/or classifying of the sequence.
|
地址 |
Caxias do Sul BR |