| 摘要 |
<p><P>PROBLEM TO BE SOLVED: To precisely determine a low-regularity and short domain in consideration of a protein domain of negative event. <P>SOLUTION: In a domain determination method, the domain of an amino acid is detected by computing a score corresponding to the composition of the domain from a known amino acid sequence database by using a hidden Markov model preliminarily constructed. Further, the detected domain is classified into a positive or negative event indicating whether the result of the detection is right from two shreshold values concomitant with the hidden Markov model and the score. The two shreshold values represent the upper limit of the score of a wrong domain and the lower limit of the score of a right domain, respectively. Then, the amino acid sequence of each domain belonging to a positive/negative event is converted to a numerical sequence by using the hidden Markov model and its characteristics are learned, and the amino acid sequence of an unknown domain is converted to a numerical sequence, and then the unknown domain is determined from the numerical sequence and the result of the learning. <P>COPYRIGHT: (C)2004,JPO&NCIPI</p> |
