Ab initio design of microbial communities from large-scale seed pools using deep learning and rapid ptimization

This repository contains the full results of our paper: Ab initio design of microbial communities from large-scale seed pools using deep learning and rapid ptimization.

Authors: Xiaoqing Jiang#, Jiaheng Hou#, Haoyu Zhang#, Jinyuan Guo, Shaohua Gu, Yulin Liao, Xinrun Yang, Peter X. Geng, Yiyan Zhou, Qian Guo, Chunhui Wang, Mo Li, Alexandre Jousset, Zhong Wei*, and Huaiqiu Zhu*

The results including:

(1) GEM.tar.gz: The eBiota-GEM dataset, containing 21,514 Genome-Scale Metabolic Models (GEMs) constructed using CarveMe based on RefSeq complete genomes.

(2) Baterial_evaluation.tar.gz: The evaluation of the ability to uptake substrates and secret productions for all 21,514 GEMs.

(3) Community_results.tar.gz: The results calculated from eBiota-GEM includes various combinations for two-bacterial consortia, covering strain IDs, substrates, products, yields, dual-bacterial growth, single-bacterial growth, co-occurrence predictions, interactions and total production.

(4) DeepCooc_files.tar.gz: The parameter files of DeepCooc, required by eBiota platform.