Fig 1.
An overview for protein identification using metaproteomic data, with metagenomic (MG) sequencing and metatranscriptomic (MT) data obtained from matched samples.
We report two novel graph traversal algorithms (Graph2Pep and Graph2Pro, highlighted in red in the figure) to extract peptides and proteins from the de Bruijn graph representation of metagenome/metatranscriptome assemblies, respectively. We note the same pipeline can be applied when only matched metagenomic or metatranscriptomic data (but not both) is available, in which the graph algorithms will be applied to the assembly graph of metagenome (or metatranscriptome).
Fig 2.
A schematic illustration of the graph traversal algorithms for extracting tryptic peptides (Graph2Pep; A) and proteins (Graph2Pro; B) from the de Bruijn graph assembly.
Table 1.
Summary of the assemblies for three data sets used in the benchmarking experiments.
Table 2.
Summary of peptide identification in wastewater datasets based on the assembly of combined metagenomic and metatranscriptomic data.
Table 3.
Improvement of protein identification by using assembly graph.
Fig 3.
Comparison of the numbers of proteins in top 20 eggNOG families receiving the most hits of proteins identified in the SD3 sample by the graph-centric approach (Graph2Pro, blue) and the conventional approach (FragGeneScan, red).
Table 4.
The additional EggNOGs protein families identified with at least 10 protein hits by the graph-centric method.
Table 5.
The number of identified enzymes involved in the Rubisco shunt.
Fig 4.
The 2-chlorobenzoate degradation pathway.
Circles represent compounds, and boxes (with EC numbers) represent enzymes. Enzymes with MS/MS data support are highlighted in purple. The figure was prepared using PathVisio [54] based on the MetaCyc’s diagrams of pathways PWY-6221 and P183-PWY.