mibPOPdb: An online database for microbial biodegradation of persistent organic pollutants

Abstract

Microbial biodegradation of persistent organic pollutants (POPs) is an attractive, ecofriendly, and cost-efficient clean-up technique for reclaiming POP-contaminated environments. In the last few decades, the number of publications documenting POP-degrading microbes, enzymes, and experimental data sets has continuously increased, necessitating the development of a dedicated web resource that catalogs consolidated information on POP-degrading microbes and tools to facilitate integrative analysis of POP degradation data sets. To address this knowledge gap, we developed the Microbial Biodegradation of Persistent Organic Pollutants Database (mibPOPdb) by accumulating microbial POP degradation information from the public domain and manually curating published scientific literature. Currently, in mibPOPdb, there are 9215 microbial strain entries, including 184 gene (sub)families, 100 enzymes, 48 biodegradation pathways, and 593 intermediate compounds identified in POP-biodegradation processes, and information on 32 toxic compounds listed under the Stockholm Convention environmental treaty. Besides the standard database functionalities, which include data searching, browsing, and retrieval of database entries, we provide a suite of bioinformatics services to facilitate comparative analysis of users' own data sets against mibPOPdb entries. Additionally, we built a Graph Neural Network-based prediction model for the biodegradability classification of chemicals. The predictive model exhibited a good biodegradability classification performance and high prediction accuracy. mibPOPdb is a free data-sharing platform designated to promote research in microbial-based biodegradation of POPs and fills a long-standing gap in environmental protection research. Database URL: http://mibpop.genome-mining.cn/.

Keywords: biodegradability classification; database; environmental pollution; microbial biodegradation; persistent organic pollutants.

Figure 1

General overview of the mibPOPdb construction, content, and web interface. mibPOPdb data were collected from published literature, biological databases, and curated scientific databases, and manually curated into three main categories. The user‐friendly online interface supports data querying, browsing, uploading new data sets, and downloading various information deposited in mibPOPdb. BBD, Biocatalysis/Biodegradation Database; CAS, Chemical Abstract Service; DB, database; mibPOPdb, Microbial Biodegradation of Persistent Organic Pollutants Database; POP, persistent organic pollutant.

Figure 2

Statistics of the collected POP‐degrading microbial data. (A) Overall distribution of microorganisms in mibPOPdb. (B) Distribution of POP‐degrading microbial strains collected from literature based on functional gene analysis. (C) Distribution of POP‐degrading microbes collected from literature based on either phylogenetic or functional screening studies. POP, persistent organic pollutant.

Figure 3

User interface of the mibPOPdb. (A) The mibPOPdb homepage. (B) Dropdown menu. Users can quickly select their POP compound from the list and organism of interest and retrieve its microbial POP degradation data sets. (C) The quick data browsing results using the dropdown menu services on the homepage. mibPOPdb, Microbial Biodegradation of Persistent Organic Pollutants Database; POP, persistent organic pollutant.

Figure 4

mibPOPdb data querying. Users can query mibPOPdb data through four paths: “Search by compound,” “Search by CAS ID,” “Search by protein and nucleotide sequence accession number,” and “Search by compound degraded.” Free texting and predictive searching are supported. (A) The search interface of mibPOPdb with compound name as the input. (B) The query's result based on “Search by the compound name” returns information on the POP compound Heptachlor. CAS, Chemical Abstract Service; mibPOPdb, Microbial Biodegradation of Persistent Organic Pollutants Database; POP, persistent organic pollutant.

Figure 5

Predicting the biodegradability of chemical compounds. Users can enter the SMILES string for their query compound. The results window shows the biodegradability probability for four representative compounds. (A) Amoxicillin, (B) musk xylene, (C) heptachlor, and (D) phenol.

Figure 6

Development of GGN‐based model for predicting chemical biodegradability. It comprises three main parts: (1) Preprocessing of data, (2) molecular graph representation of the chemical using GNNs, and (3) modifying parameters of GNNs using an optimizer based on the value of the loss function. GNN, Graph Neural Network; LSTM, long short‐term memory.