Page 42 - Read Online

Wang et al. Microbiome Res Rep 2024;3:39 https://dx.doi.org/10.20517/mrr.2024.21 Page 7 of 17

Figure 1. Experimental Overview and evaluation of bioinformatics workflow. (A) Flowchart depicting mouse fecal sample processing
workflow and MS data acquisition. (B) Flowchart depicting different bioinformatic workflows tested in this study. Histograms of the total
number of peptides and proteins identified with the different bioinformatics processing workflows as shown in (B) for (C) DDA and (D)
DIA datasets, respectively. MS: Mass spectrometry; DDA: data-dependent acquisition; DIA: data-independent acquisition.

A major challenge for DIA metaproteomic data analysis is that bioinformatics tools, such as DIA-NN,
cannot handle large databases. Currently, most workflows utilize DDA data of the same or representative
samples to generate a spectral library from the original protein databases. We therefore first evaluated the
use of two widely used database search engines for DDA data, pFind (open search) and MSFragger (closed
search with split databases), to generate reduced databases from the mouse gut microbial gene catalog
database [Figure 1B]. The pFind open search for all 30 DDA data files identified 91,524 peptides
corresponding to 15,011 protein groups in total, while the MSFragger search for the 30 DDA data files
identified 54,093 peptides corresponding to 15,077 protein groups [Figure 1C]. DIA-NN searches with
spectral library or library-free modes from either pFind or MSFragger reduced databases were then