RiboMicrobe: An Integrated Translatome Atlas for Microorganism

Area of Science:

• Microbial genomics and transcriptomics
• Computational biology and bioinformatics
• Molecular biology and genetics

Background:

• Ribosome profiling (Ribo-seq) is crucial for studying translation dynamics and identifying small open reading frames (sORFs).
• Existing translatome databases lack comprehensive coverage of prokaryotic Ribo-seq data and sORF analysis.
• There is a need for specialized resources to facilitate the analysis of translation in prokaryotes.

Purpose of the Study:

• To develop a comprehensive database, RiboMicrobe, for Ribo-seq data analysis in prokaryotic microorganisms.
• To introduce novel deep learning-based models (sORFPredRibo and sORFPred) for accurate sORF prediction.
• To provide integrated bioinformatics tools for exploring, predicting, and comparing translational data.

Main Methods:

• Compilation of a large-scale database integrating Ribo-seq, RNA-seq, and proteome datasets from prokaryotes and viruses.
• Development of transformer-based deep learning models for sORF prediction.
• Implementation of a web interface with visualization, prediction, and comparative analysis tools.

Main Results:

• RiboMicrobe database now hosts 891 Ribo-seq, 369 RNA-seq, and 62 proteome datasets from 36 prokaryotes and 2 viruses.
• Two novel sORF prediction models, sORFPredRibo and sORFPred, have been developed and integrated.
• A suite of bioinformatics tools for data exploration, sORF prediction, and functional investigation is available.

Conclusions:

• RiboMicrobe provides a valuable, centralized resource for prokaryotic translatomics research.
• The integrated deep learning models enhance the accuracy of sORF identification in prokaryotes.
• This resource is expected to significantly advance the field of microbial translational research and sORF annotation.