一种基于新的向量空间模型的谱库检索算法

Novel Vector Space Model and Algorithm for Search of Mass Spectral Library

在谱库检索中,可以通过建立向量空间模型,将谱图数据表示为向量空间中的向量,计算向量之间的相似性即可得到谱图的相似程度。本文提出一种新的向量空间模型,并基于新的模型提出谱库检索算法,实验发现,新的模型不仅降低了向量空间的维度,减少了计算量,并且谱图之间的相似性计算结果优于现有的基于空间向量模型的算法,检索结果与NIST05完全相同。

A novel algorithm was developed for search of the mass spectral library, on the basis of newly-sim- plified vector space model. In searching the mass spectral library, the standard mass spectra can be described as a vector in the vector space model. The mass spectrum of anunknown compound can be routinely identified by search- ing the mass spectral library to single out the standard mass spectrum that bears the best similarity. The original work involved a significant reduction of the dimensions of the vector space. When it comes to the search of mass spectral library, the novel algorithm outperforms the conventional one because of shorter computation time, faster speed, smaller storage space and better similarity. A round robin test for identification of some compounds with mo- lecular weight of 500amu in the new and NIST05 algorithms was performed. The searching results turned out to be exactly the same.