Compound identification using random projection for gas chromatography-mass spectrometry data

Abstract

In general, compound identification through library searching is performed on original mass spectral space by using some developed similarity measure. In this paper, the original mass spectral space was transformed into binary space by random projection. The hamming distance between query and reference the vector of binary space are calculated. The Mass Spectral Library 2005 (NIST05) main library is used as reference database and the replicate library is used as query data. With the number of binary digits increasing, the accuracy of compound identification is also increased. When the number set as 2076 bits, random projection achieve better identification performance than corresponding three similarity measures.