Bioinformatics in silico by the Unification of Symobols and Patterns

符号和模式统一的计算机生物信息学

基本信息

批准号：
17200016
负责人：
MATSUYAMA Yasuo
金额：
$ 29.2万
依托单位：
Waseda University
依托单位国家：
日本
项目类别：
Grant-in-Aid for Scientific Research (A)
财政年份：
2005
资助国家：
日本
起止时间：
2005 至 2007
项目状态：
已结题

来源：
https://kaken.nii.ac.jp/en/grant/KAKENHI-PROJECT-17200016/
关键词：
bioinformatics soft computing DNA sequence analysis promoter transcription start site prediction protein function analysis tertiary structure prediction proteome information processing バイオインフォマティクスゲノム情報処理

项目摘要

This project was started towards the development of computational intelligence algorithms for finding soft patterns existing in DNA and amino acid sequences. The main methodology is in Aim. Wet biologists are included in this group so that overly abstract problems are suppressed. The unification between compute-based information scientists and test-tube-based life scientists still requires time, however, a steady step towards such collaboration was enhanced by this project with the following results :(1) Prediction methods fir the transcription start site were established. On human .genome which is a representative of eukaryotes, a combination of the spectrum kernel, hidden Markov models, and FFT integrated by a support vector machine was presented. This mechanism yielded a top class ROC curves. On the prediction of E.coli which is a representative of prokaryotes, a combination of the independent component analysis and a support vector machine revealed the best prediction performance to date.(2) Anew effective algorithm on the multiple sequence alignment was developed. This new method suppresses the appearance of multiple gaps in the same column. The gap extension can be regulated by piecewise linear penalties. The total algorithm is realized as the software named PRIME. The PRIME showed better performances than ClustalW and T-Coffee in the sense of resulting alignments and computational speed.(3) The wet biology team hind an evidence on Rad5l which repairs cut double strands of DNA. The binding site of Rad51 is altered in breast cancer patients.As was explained above, this research brought about fruitful results on post genome topics : The prediction of promoters and transcription start sites, a new multiple sequence alignment method leading to tertiary structure prediction, and a cancer property caused by protein functions.

该项目旨在开发计算智能算法，用于查找 DNA 和氨基酸序列中存在的软模式。主要方法论在 Aim 中。湿生物学家也被纳入这一群体，从而抑制了过于抽象的问题。基于计算的信息科学家和基于试管的生命科学家之间的统一仍然需要时间，但是，该项目为这种合作迈出了坚实的一步，取得了以下成果：（1）建立了转录起始位点的预测方法。在作为真核生物代表的人类基因组上，提出了频谱核、隐马尔可夫模型和支持向量机集成的FFT的组合。这种机制产生了顶级的 ROC 曲线。在对原核生物代表大肠杆菌的预测中，独立成分分析和支持向量机的结合显示了迄今为止最好的预测性能。(2)开发了一种新的多序列比对的有效算法。这种新方法抑制了同一列中出现多个间隙。间隙扩展可以通过分段线性惩罚来调节。整个算法以PRIME软件的形式实现。在结果比对和计算速度方面，PRIME 显示出比 ClustalW 和 T-Coffee 更好的性能。(3) 湿生物学团队提供了 Rad5l 修复 DNA 双链的证据。 Rad51的结合位点在乳腺癌患者中发生改变。如上所述，这项研究在后基因组主题上取得了丰硕的成果：启动子和转录起始位点的预测，一种新的多序列比对方法，导致三级结构预测，以及由蛋白质功能引起的癌症特性。