Abstract:
A method for determining monoisotopic masses by determining isotopic clusters is provided to ensure a high accuracy and high processing speed in determining the position of a real monoisotopic peak even when a polypeptide mixture has a composition away from a general amino acid composition, and complex peaks of isotopic clusters appear in the mass spectrum. A method for determining possibility models of isotopic clusters for determining monoisotopic masses of isotopic clusters comprises the steps of: approximating each peak intensity (Ik), ratio of two peak intensities(Ik+1/Ik) and multiplication of two ratios(Ik-1Ik_1/I^2) obtained from three peaks in the isotopic clusters to a possibility formula; and calculating the highest, lowest and average functions(Rmax(k, M), Rmin(k, M), Ravg(k, M)) of the k^th ratio of a polypeptide with a mass of M, and the highest, lowest and average functions (RPmin(k, M), RPmax(k, M), RPavg(k, M)) of the multiplication of the k^th ratio of the polypeptide with a mass of M by using the possibility formula.
Abstract:
PURPOSE: A high speed multiple-unrestricted modification searching method through a mass spectrometry and a device thereof are provided to use dynamic programming algorithm based on a multi sequence tag from each a mass spectrometry spectrum according to modification alignment, thereby determining modifications about a database peptide. CONSTITUTION: MS/MS spectrum information is inputted from a mass spectrometry. A plurality of amino acid ranking tags is induced. Each of ranking tags is arranged about candidate peptides including the corresponding ranking tags in a database. Modification within the candidate peptide is identified without a limit of the number of modifications by every peptide through dynamic programming limiting a moving route according to location information of tags in a matrix. Each of the ranking tags are arranged for arrangement between ranking of the spectrum information and the candidate peptides in the matrix.
Abstract:
PURPOSE: A method for generating a keyword of a document by using ontology information and a device thereof are provided to generate the keyword based on the ontology information set in a word of the document, thereby generating a keyword for the document which the keyword is not provided. CONSTITUTION: An ontology setting unit(210) sets an ontology word in a document by using words included in the document. A candidate keyword extraction unit(230) extracts a candidate keyword which is composed of words of the document. A candidate keyword extension unit(240) extends the candidate keyword by using an association relation between the ontology word and the candidate keyword. A final keyword selection unit(250) sets ranking of the candidate keyword by using a machine learning algorithm and selects the candidate keyword as the final keyword. [Reference numerals] (210) Ontology setting unit; (230) Candidate keyword extraction unit; (240) Candidate keyword extension unit; (250) Final keyword selection unit; (260) Rank resetting unit; (AA) Document input; (BB) Keyword output
Abstract:
본 발명은 온톨로지 정보를 이용한 문서 주제어 생성 방법 및 그 장치에 관한 것으로, 본 발명의 일 실시예에 따른 문서 주제어 생성 방법은 문서에 포함된 단어를 이용하여 상기 문서에 온톨로지 용어를 설정하는 단계와, 상기 문서에 포함된 적어도 하나의 단어로 구성되는 후보 주제어를 추출하는 단계와, 상기 온톨로지 용어와 상기 후보 주제어 간의 연관관계를 이용하여 상기 후보 주제어를 확장하는 단계와, 상기 확장된 후보 주제어를 기계학습 알고리즘을 이용하여 순위를 설정하고, 미리 설정된 순위 이상의 후보 주제어를 최종 주제어로 선택하는 단계를 포함함으로써, 주제어가 제시되지 않은 문서에 대한 주제어를 생성할 수 있다.
Abstract:
상관관계 합의 선형시간 계산 방법을 개시한다. 상관관계 합의 선형시간 계산 방법은 두 신호의 상관관계(cross correlation)의 합을 계산하는 상관관계 합의 계산 방법에 있어서, 실험 대상이 되는 상대신호(target signal)를 제1 입력신호로 입력받고 기준신호(base signal)를 제2 입력신호로 입력받는 단계; 및, 상기 제1 입력신호와 제2 입력신호 중 어느 한 입력신호의 각 원소를 더하는 덧셈 연산을 먼저 수행한 후 상기 덧셈 연산의 결과를 다른 입력신호의 각 원소에 곱하는 곱셈 연산을 수행하는 연산 순서에 따라 상관관계의 합을 계산하는 단계를 포함할 수 있다. 프로테오믹스(proteomics), 단백질 동정, 탠덤 질량 분석(tandem mass spectrometry), 시퀘스트(SEQUEST), FFT(Fast Fourier Transform), 상관관계(cross-correlation) 연산, 유사도, 선형 시간
Abstract:
PURPOSE: A method for calculating linear time of cross-correlation is provided to drastically shorten data processing time for correlation sum. CONSTITUTION: An experiment spectrum is obtained(S210). A fixed number of candidate amino acid ranks is selected among selected peptide amino acid ranks in a peptide rank database(S220). A virtual spectrum is generated through the selected candidate amino acid rank(S230). Similarity between the experiment spectrum and the virtual spectrum is calculated through a linear time operation method of a cross-correlation sum. The peptide amino acid rank of the experiment spectrum is determined(S240).