Machine learning 'improves MS analysis of biologically active substances'

Analysing biologically active substances using tandem mass spectrometry (MS) can be improved using a new machine learning protocol, according to scientists.

Writing in BMC Bioinformatics, a periodical dedicated to the latest statistical and computational methods for data analysis, a team from the University of Illinois at Chicago explain how they were able to achieve a six per cent improvement over previous protocols.

They write: "We present a machine learning based protocol for the identification of correct peptide-spectrum matches from Sequest database search results."

Within their programming is the ability to define multiple rules on an additive basis, allowing "expert rule of thumb" approaches to be emulated.

In turn, this means the system is able to carry out its calculations without what the team calls the "black box notion".

Tandem MS is particularly used in large-scale studies of biologically active substances where proteins must be characterised based on their matching with peptide-spectrum database records.