Computerised gas-liquid chromatography of bacterial cellular fatty acids in analysis of bacterial mixtures

Gas-liquid chromatography (GLC) of bacterial cellular fatty acids (CFA) was tested as a method to obtain species-specific information directly from multimicrobial samples. The CFA spectra of 145 mixtures containing 3, 4, 6 or 8 microbial species were produced by GLC. The idea of direct GLC of multibacterial samples is based on the assumption that the CFA composition of a sample reflects all the bacteria present in the sample. Correlation and cluster analyses of the spectra were performed and the results were presented as dendrograms. When all the CFAs in the samples were employed in the analysis the microbial species in the samples could not be identified. When repeated correlation and cluster analyses were performed employing at a time only the CFAs specific for one microbial species present in the mixtures (index species), the identification of the samples containing the index species was invariably correct in the mixtures containing 3 microbial species. In the mixtures of 4 species the sensitivity of detection was 98.9% and the specificity 94%. When various statistical approaches were used, the identification of the samples with the index species was also possible in the mixtures with 6 species, although the relative similarity of the CFA spectra of certain index species emerged then as a limiting factor in the identification. We conclude that species-specific information could be obtained from all the mixtures by employing species-specific CFAs in the analyses. The identification of a species present in a mixture was successful when the amount of the species was greater than or equal to 25% of the bacterial mass.