MPA_Pathway_Tool. it consists of two parts, called ‘Pathway-Creator’ and ‘Pathway-Calculator’. The ‘Pathway-Creator’ enables an easy set-up of user-defined pathways with specific taxonomic constraints. The ‘Pathway-Calculator’ automatically maps microbial community data from multiple measurements on selected pathways and visualizes the results. The MPA_Pathway_Tool is implemented in Java and ReactJS.
There are several different approaches for analyzing microbial communities, focusing on the entirety of the genes (metagenomics), transcripts (metatranscriptomics), or proteins (metaproteomics). Whereas metagenomics reveals only the presence of genes, and metatranscriptomics only the gene expression, metaproteomics indicates actual protein expression 
. On the basis of the protein expression levels, a microbial communities’ phenotype can be linked with specific environmental conditions, process parameters, or diseases 
. Due to the complexity and amount of multi-omics data, comprehensive bioinformatic workflows were developed for the data evaluation 
. For example, the MetaProteomeAnalyzer (MPA) enables the analysis and and interpretation of metaproteomic data sets. It offers a free, open-source, end user-oriented complete pipeline from peak lists to taxonomic and functional result evaluation. Among others, the MPA links identified proteins to functional categories (e.g., biological keywords) and the KEGG pathways 
. In addition to the KEGG pathway system 
, several other pathway collection and mapping tools such as Reactome 
, Escher 
, and Pathway Tools 
exist, supporting the data analysis of omics-datasets. For more details about the assignment of genes and proteins to functions and pathways, please refer to Mao et al. (2005) 
However, due to the microbial metabolism’s versatility and constantly newly discovered biological pathways 
, linkage to standard pathway maps is insufficient for many microbial community studies. Therefore, new tools are required that are tailored for microbial community studies.
, while the Wood–Ljungdahl pathway is carried out mainly by bacteria 
. (vi) The tool should be independent of operating systems so that nearly everyone can use the tool, favoring an implementation as a web application.
2. The ‘Pathway-Creator’ Enables Users to Define Their Own Metabolic Pathways
The first part of the MPA_Pathway_Tool, the ‘Pathway-Creator’ (Figure 1), enables the creation of user-defined pathways by adding reactions iteratively and linking omics data to this pathway. The menu (left side) allows for uploading of experimental data and pathways (such as CSV, JSON, and SBML), adding new user-defined reactions or reactions from KEGG, and downloading created pathways (such as CSV, SBML, JSON, and Scalable Vector Graphics (SVG)) and mapped data (such as CSV). The right side visualizes the created pathway. Circular-shaped nodes (metabolites; in KEGG referred to as compounds) and diamond-shaped nodes (reactions) are connected by arrows displaying the direction of each reaction. After a sample of previously uploaded data is selected by clicking on the respective button (bottom side of the tool), reaction nodes are colored dependent on their abundance in the sample. Information about abundances in all samples for a specific reaction is available as a heatmap by clicking on the respective reaction node.
Screenshot of the ‘Pathway-Creator’ of the MPA_Pathway_Tool (details in Section 2.1
3. The ‘Pathway-Calculator’ Enables Automated Mapping of Experimental Data on Multiple Metabolic Pathways
The ‘Pathway-Calculator’ (Figure 2
) consists of two upload zones, one for experimental data and another for multiple pathway files (such as CSV, JSON, or SBML). It performs mapping (details in Section 3.3
) of experimental data on uploaded pathways. After experimental data are mapped on uploaded pathways, the resulting table can be downloaded as CSV. Furthermore, a list with all unmatched features (e.g., proteins) can be exported. Details about the taxonomic structure of data mapped on analyzed pathways can be found under ‘See Details’. Finally, we also provide metadata about the mapping containing among others mapping time, used experimental data, and pathways. It fulfills the FAIR (findable, accessible, interoperable, and reusable) principles 
and highlights the importance of emerging metadata standards 
Figure 2. Screenshot of the ‘Pathway-Calculator’ of the MPA_Pathway_Tool (details in Section 2.2).
Finally, we tested the performance of the ‘Pathway-Calculator’ by uploading experimental data with different file sizes (10,000 proteins, 100,000 proteins, and 1,000,000 proteins with 44 samples per test) and a different number of pathways on a local desktop computer (AMD Ryzen 5 3600, 16 GB DDR4 RAM 3000 MHz, Chrome Browser version 89.0.4389.90). We utilized 1, 10, and 100 copies of the user-defined Wood–Ljungdahl pathway used in Heyer et al. 
(Biogas plant (BGP)-Wood–Ljungdahl pathway) for the performance test. Each test with 10,000 and 100,000 proteins finished within 1 min, indicating a good performance for most files. Tests with 1,000,000 proteins took longer (up to 12 min) caused by higher upload times and high requirements on memory and on CPU performance (Table 1
Table 1. Performance test for the ‘Pathway-Calculator’ (details in Section 2.2).
4. Tool Validation with Experimental Data
Validation of the MPA_Pathway_Tool was carried out by comparing our pathway assignment against a manual pathway assignment of a previous metaproteomics study regarding biogas plants 
. Additionally, we assigned the proteins to the standard KEGG pathways (in KEGG referred to as KEGG MODULEs) to illuminate the demand for user-specific pathways.
For the evaluation, we created six pathway maps (BGP-hydrogenotrophic methanogenesis, BGP-acetoclastic methanogenesis, BGP-Wood–Ljungdahl pathway, KEGG-hydrogenotrophic methanogenesis, KEGG-acetoclastic methanogenesis, and KEGG-Wood–Ljungdahl pathway). The user-defined pathways (BGP; Figures S3–S5
) and the KEGG pathways (Figures S6–S8
) were created with the ‘Pathway Creator’ (further details in Supplementary File S17
). For each pathway, a taxonomic classification was added (Table 2
). According to the publication of the dataset, we decided to exclude Archaea from the Wood–Ljungdahl pathway and include only Archaea for the hydrogenotrophic and acetoclastic methanogenesis 
. Subsequently, the experimental metaproteomics data (Supplementary File S9
) were mapped on each pathway using multiple pathways (Figure 3
; Supplementary File S1
) and single pathway mapping (Figure 4
; Supplementary File S10–S15
). Unmatched proteins were downloaded as CSV (Supplementary File S2
Comparison of experimental data mapped on each pathway (Wood–Ljungdahl pathway, acetoclastic methanogenesis, and hydrogenotrophic methanogenesis) with published data 
. Summed spectral counts plotted against all biogas samples are blue and yellow for user-defined pathways and black for KEGG MODULEs. The results were obtained from the ‘Pathway-Calculator’ (yellow and black) and published data (blue). In particular, acetoclastic methanogenesis showed higher spectral counts in most samples, indicating the occurrence of unspecific reactions.
Summed spectral counts of each reaction in the Wood–Ljungdahl pathway, acetoclastic methanogenesis, and hydrogenotrophic methanogenesis. Results from published data 
are visualized in blue, results from the ‘Pathway-Calculator’ that are user-defined are shown in yellow, and results from the ‘Pathway-Calculator’ for KEGG MODULEs are shown in black. All KEGG MODULEs possess additional unspecific reactions. KEGG-acetoclastic methanogenesis and KEGG-hydrogenotrophic methanogenesis possess three additional reactions (18.104.22.168, 22.214.171.124., and 126.96.36.199) that catalyze the last steps of the methanogenesis (see Supplementary File S19).
Table 2. Taxonomic classification of each pathway.
||Added Taxonomic Requirement
|BGP-hydrogenotrophic methanogenesis (user-defined)
|KEGG-hydrogenotrophic methanogenesis (KEGG-MODULE)
|BGP-acetoclastic methanogenesis (user-defined)
|KEGG-acetoclastic methanogenesis (KEGG-MODULE)
|BGP-Wood–Ljungdahl pathway (user-defined)
||all except Archaea
|KEGG-Wood–Ljungdahl pathway (KEGG-MODULE)
||all except Archaea
The summed spectral counts mapped on each user-defined pathway were slightly greater than in the published experimental data, indicating a small information loss in the previous publication (Supplementary File S18).
Furthermore, we compared the results of each BGP pathway with the corresponding KEGG pathway. We observed higher spectral counts for the created KEGG MODULEs compared to the respective BGP pathways (Figure 3). Higher spectral counts in KEGG MODULEs are caused by an integration of more reactions in the KEGG MODULEs than in the BGP pathways (Figure 4). For example, KEGG-acetoclastic methanogenesis and KEGG-hydrogenotrophic methanogenesis possess three additional reactions (188.8.131.52, 184.108.40.206., and 220.127.116.11) that catalyze the last steps of the methanogenesis. BGP pathways include only pathway-specific reactions to prevent wrong pathway identification.
5. Conclusion: The MPA_Pathway_Tool Provides An Easy and Fast Option to Set up Multiple Pathways
We successfully implemented the pathway tool to meet all of our six defined criteria: (i) The MPA_Pathway_Tool provides an easy and fast setup of multiple pathways. Multiple reactions can be imported using various options, e.g., import by EC numbers, import of a KEGG MODULE, or import of entire SBML files. (ii) A further modification of the generated pathways is possible by deleting reactions and adding new reactions (from the KEGG database and user-defined reactions). (iii) As interchange formats, JSON, CSV, and SBML were implemented. (iv) Experimental data from metaproteomics, metatranscriptomics, and metagenomics studies can be automatically mapped on single pathways (‘Pathway-Creator’) and multiple pathways (‘Pathway-Calculator’). The results of pathway mapping can also be exported as CSV. (v) The mapping algorithm includes a taxonomic filter that was successfully applied by comparing our results with published experimental data for the stated pathways. (vi) The MPA_Pathway_Tool was implemented as a stand-alone web application to guarantee independence from users’ operating systems.