摘要
解链温度是目前评价DNA序列热力学稳定性的一个主要的参数,目前,生物工程中常用的各种预测方法都存在某些序列的误差偏大的缺点,因此难以满足像DNA计算这种大量DNA序列进行各种生化反应的计算过程的要求。本文以DNA序列的邻近法参数为基础,根据DNA序列热稳定性特点提出了DNA序列的编码方案,并建立了基于特征自组织网络的解链温度的预测模型。计算结果表明,DNA序列的解链温度的误差可以达到±5℃的范围。
Methods for Prediction of DNA duplex thermodynamics from sequence and DNA concentration are based on the nearest-neighbor (n-n) mode, but sequence dependent interactions in DNA duplexes may extend over distances beyond. In order to calculate Melting Temperature (Tm) in a more reasonable way, we use the SOFM of kohonen neural-networks to extract wild field features from DNA sequence, and a back propagation network is employed to predict Tm of DNA duplexes with these features. This method considers interactions of DNA sequence beyond nearest-neighbor and the precision of the Prediction result is in±5℃error band.
出处
《计算机与应用化学》
CAS
CSCD
北大核心
2006年第6期515-518,共4页
Computers and Applied Chemistry
基金
The supported by the Chinese National Natural Science Foundation(No.60274026,60373089).