3‐Siderophore Classification
Based on digitized siderophores in our database, we try to classify known siderophores.
Refer to Enzyme Commission number and the Transport Classification system, we introduce similar nomenclature for siderophore classification.
In general, our siderophore classification is based on structure similarity.
To aviod artificial threshold in clustering, we propose a novel method to classify cluster of siderophores.
The main idea is that different siderophore clusters will emerge when they are put in a larger chemical space.
In our study, we use COCONUT natrual product with 407,270 molecules as the chemical space background.
Displaying 25 clusters of 649 siderophores in the COCONUT database by TMAP
You can the script for this figure in here.
649 siderophores were split into 25 distinct clusters.
The visualization of large siderophore clusters in SIDERITE.
A. A network of siderophores connected by chemical similarities. Each node in the network corresponds to a siderophore molecule, and the nodes are linked to their most similar neighbors, forming a minimum spanning tree(57) (described in the method section). Nodes are colored by their cluster IDs.
B. For the four largest clusters, example structures are provided, and the functional groups of the siderophores are colored according to their types. Siderophores are circled by rounded rectangles to show their cluster IDs (same color codes as (A)).
You can the script for this figure in here.
Within each cluster, there are common features of functional groups or biosynthetic types.
Cluster 1 (201, 30.97%) includes siderophores with phenyl ring structures in the functional groups such as catecholate, phenolate, hydroxyphenyloxazoline, and hydroxyphenylthiazoline. Siderophores in the cluster 1 are synthesized by both NRPS and NIS.
Cluster 2 (197, 30.35%) only includes siderophores produced by NRPS pathways except Albomycins by the hybrid NRPS/NIS pathway. Albomycins are naturally occurring sideromycins (siderophore–antibiotic conjugates) produced by some streptomycetes, and their siderophore parts are synthesized by NRPS.
From the perspective of functional groups, most siderophores in the cluster 2 contain hydroxamate of (92.39%, 182/197), and many also contain alpha-hydroxycarboxylate (37.06%, 73/197).
Cluster 3 (103, 15.87%) is all NIS siderophores. Like cluster 1, most of them contain hydroxamate (90.29%, 93/103), and many also contain alpha-hydroxycarboxylate (33.01%, 34/103). The sources of alpha-hydroxycarboxylate are mostly citrate.
Cluster 4 (79, 12.17%) is NRPS siderophores with chromophores such as pyoverdines (93.67%, 74/79). Other small clusters all are located on the edge of four large clusters (Fig 2A).
Clusters only provide an initial classification relative to other natural product molecules.
Further, we defined groups within each of the 25 clusters by their structure similarity coefficient (Dice coefficient, threshold 0.6), and obtained 102 groups in total.
Each group is named by its cluster ID x and its group ID y within this cluster. Accordingly, each siderophore is assigned a unique ID x.y.z, where z stands for the z-th record within the group.
For example, enterobactin has the ID 1.22.2. In this ID, “1” means the 1st large cluster. “22” means the 22nd group within the 1st large cluster. “2” means the 2nd record in the 1.22 cluster.
In the future, newly discovered siderophores will also be assigned a unique id when being incorporated into the SIDERITE database.