Gut microbiome is perhaps the most complex microbial community, consisting of approximately 10¹³ microbial cells spanning hundreds to thousands of different species. Recent researches have revealed the myriad associations between gut microbiome and human physiology, health and diseases. Metaproteomics is a newly developed proteomics technology aiming at the high-throughput identification and quantification of the entire protein complement of a microbial community. It is capable of characterizing the functional profile of the gut microbiome in real time and establishing a more accurate correlation with the host phenotypes or environmental factors. However, the extremely high diversity and complexity of the gut microbiome make it difficult to identify and quantify proteins from metaproteo...[ Read more ]

Gut microbiome is perhaps the most complex microbial community, consisting of approximately 10¹³ microbial cells spanning hundreds to thousands of different species. Recent researches have revealed the myriad associations between gut microbiome and human physiology, health and diseases. Metaproteomics is a newly developed proteomics technology aiming at the high-throughput identification and quantification of the entire protein complement of a microbial community. It is capable of characterizing the functional profile of the gut microbiome in real time and establishing a more accurate correlation with the host phenotypes or environmental factors. However, the extremely high diversity and complexity of the gut microbiome make it difficult to identify and quantify proteins from metaproteomics samples, impeding the comparative analysis across samples. This thesis set out to develop a metaproteomics method, circumventing the protein identifications and performing comparisons between samples on the basis of the MS/MS spectra acquired. We employed the spectrum clustering algorithm, SpectraST, to construct a consensus spectra library of all samples, and then determined the similarities and differences between samples by quantifying the contribution they made to the consensus spectra library. We successfully grouped 48 metaproteomics samples into 4 distinct clusters, and verified them using metagenomics analysis. We then integrated database search, de novo sequencing, and open search, trying to identify the consensus spectra showing significant differences between clusters. A reconciliation scheme was proposed to coordinate conflicting answers from different identification methods and from the corresponding raw spectra of the consensus spectra. In contrast to the conventional proteomics workflow, our method carried out unsupervised clustering and comparison between samples priori to protein identification, whereby individual variance can be mitigated.