The annotations of MGCs and BGCs are valuable for identifying metabolic pathways. This helps to provide more evidence for interactions between microbes and their associated primary or secondary metabolites. MGCs are predicted by gutSMASH [Ref: Nucleic Acids Research, 49.W1 (2021), W263–W270] while BGCs are generated by antiSMASH [Ref: Nucleic Acids Research, 51.W1 (2023), W46–W50]. Using hidden Markov models (pHMMs), these two algorithms perform comparative genomic analysis with two customized databases, KnownClusterBlast and ClusterBlast, to MGCs or BGCs from microbial genomes. Due to the lack of strain-level reference genomes, species-level reference genomes of microbes within each genus from MASI database are downloaded from NCBI for analysis. Finally, we display the related compounds as substrates or products of each MGC or BGC with a similarity percentage of 100% calculated by KnownClusterBlast.

You can click on different parts of the human body or select in the table to see the microbe abundance in that specific region.

To obtain the reactions associated with therapeutic substances, we followed a multi-step process:
Downloading Reconstructions: We started by downloading microbial genome-scale metabolic reconstructions from the AGORA2 [Ref: Nature Biotechnology, 41 (2023) 1320?331] database.
Identifying Drug-Associated Reactions: Next, we extracted all reactions that are linked to therapeutic substances from these reconstructions. This involved filtering and identifying reactions specifically related to drug metabolism and transport.
Linking Reaction to Microbes: Utilizing the identified reaction related genes (UidA, Tdc etc.), we machted the corresponding drug-associated reactions to existing microbes in the reconstructions in AGORA2. We could link the presence of these genes in different microbes to the potential for those microbes to carry out the corresponding drug-related reactions.
Putative Drug Reactions: As a result, the drug reactions identified in this manner are putative, meaning they are inferred based on the presence of specific gene sequences. This provides a hypothetical but informed prediction of the microbial capability to interact with therapeutic substances.

To obtain the reactions associated with dietary substances, we followed a multi-step process:
Downloading Reconstructions: We started by downloading microbial genome-scale metabolic reconstructions from the AGREDA [Ref:Nature Communications, 12 (2021) 4728] database.
Identifying Diet-Associated Reactions: Next, we extracted all reactions that are linked to dietary substances from these reconstructions. This involved filtering and identifying reactions specifically related to dietary substance metabolism and transport.
Linking Reactions to Microbes: Using the identified related genes (e.g., UidA, Tdc) for each drug metabolite reaction, we matched these reactions to microbes possessing the corresponding genes. This allowed us to link the presence of these genes in different microbes to their potential for carrying out the associated drug-related reactions.
Putative Drug Reactions: As a result, the diet reactions identified in this manner are putative, meaning they are inferred based on the presence of specific gene sequences. This provides a hypothetical but informed prediction of the microbial capability to interact with dietary substances.

ⓘ What does QS language mean?

Quorum sensing (QS) is a system of stimulus and response correlated with population density. Many microbes use quorum sensing to regulate gene expression based on the size of their population.

To gather potential microbe-to-microbe interactions based on quorum sensing, we utilized several sources and techniques, including:

• QSHGM Database [Ref: Nat Commun 13, 3079 (2022)]: This database provides extensive information on quorum sensing and quorum quenching genes in microbes based on machine learning algorithms.
• QSDB Database [Ref: Database (Oxford). 2021 Nov 26;2021(2021):baab071]: The Quorum Sensing Database (QSDB) contains curated information on microbial signaling pathways, including interactions mediated by specific quorum sensing molecules.
• Manual Data Collection: In addition to database resources, we manually collected data from scientific literature, research articles, and experimental studies.

By combining these data sources, we created a putative species-level network to visualize potential microbe-to-microbe interactions.

ⓘ User Guideline:

1. Understanding the Nodes and Edges:
· Nodes:
Nodes can be categorized into two types: QS languages and associated microbes (species level):
🖈QS language nodes: These are the QS languages associated with the microbe you have chosen.
🖈Microbe nodes: These represent the corresponding species. Microbes with different numbers of associated QS languages are displayed in varied colors. Due to the force field algorithm, microbe nodes connected with more QS languages tend to gather at the center of the plot, while those with fewer QS languages scatter outside.
· Edges:
Each edge is defined as follows: if the chosen microbe is reported or predicted to be associated with a known QS language, an edge will be established between the QS node and the microbe. All strain-level connections are merged into species-level connections to show potential interactions. For more details of related strains under one QS language, please refer to the table below the network.

2. Exploring Interactions:
· Drag: Feel free to drag and explore the area that interests you.
· Zoom in and out: Use the mouse wheel to zoom in and out if you wish to see the nodes and edges in larger size.
· Hide or show: Click the labels of microbes on the top of the plot with different connected QS languages to hide or show them.