Dataset for Neural Network

This is the dataset for the manuscript named with "Neural Network Establish Co-Occurrence Links between Transformation Products of Contaminants and Soil Microbiomes". Quantitative data of HRMS were provided here. The abstract for the manuscript is as following: It is still challenging for ecologists and environmentalists to identify which microorganisms are carrying out specific metabolic processes in the natural environment, even though stable isotope probing (e.g., DNA-SIP) could link degraders and their substrates. As a new strategy, we combined the use of a network-based algorithm, MMvec, and our developed 2H-labeled Stable Isotope-Assisted Metabolomics pipeline (2H-SIAM) to discover links between transformation products (TPs) of the contaminant and microbes in soils. Abiotic stresses were firstly used to constitute the assembly of soil microbiomes, characterized by 16S rRNA gene sequencing. Pyrene and pyrene-d10 were added into soils for biodegradation, and 2H-SIAM was used to obtain TPs of pyrene. Then, MMvec was used to establish a co-occurrence network between TPs and microbiomes. The results confirmed the role of Pseudomonas and Phenylobacterium in the oxidation, mineralization, and methylation of pyrene. Sphingomonas and phylum Acidobacteria contributed to the oxidation of pyrene. The obtained co-occurrence network was markedly following the reports studied by DNA-SIP, indicating the performance and reliability of the co-occurrence network. In summary, we firstly depict the links between TPs and microbes in the environment matrix, which exhibits unique advantages comparing to the other isotope-based approaches..
The installation of the MMvec please refer to https://github.com/biocore/mmvec.
In this study, the MMvec was carried out in qiime2-2020.6 platform, and the MMvec was carried out with the codes provided in a .doc document. Additionally, two necessary documents, "lcms_nt.txt" and "otus_nt.txt", for the evaluation of the MMvec are provided here.