Rapid and quantitative analysis of bioprocesses using pyrolysis mass spectrometry and neural networks: application to indole production

In pure form indole, when subjected to pyrolysis mass spectrometry @MS), gave a pattern of peaks at m/z 117, 90,89 and a murmur at 63. Significant differences in the magnitudes of these peaks were observed between strains of Escherichia coli which were grown on nutrient agar and which differed solely in whether a transposon had been inserted into the tryptophanase gene or elsewhere within the genome. We applied artificial neural networks (ANNs) to the deconvolution of pyrolysis mass spectra. The combination of ANNs and PyMS was able quantitatively to detect the component indole when a single strain of E. coli, containing the tryptophanase gene, was grown on a minimal supplemented salts medium incorporating various amount of tryptophan, in the range O-253 mg/l. This approach constitutes a novel, powerful and interesting technology for the analysis of the concentrations of appropriate substrates, metabolites and products in chemical and bioprocesses generally.