Devising Isolation Forest-Based Method to Investigate the sRNAome of Mycobacterium tuberculosis Using sRNA-seq Data

Area of Science:

• Genomics and Molecular Biology
• Bacterial Pathogenesis
• Bioinformatics and Computational Biology

Background:

• Small non-coding RNAs (sRNAs) are crucial regulators of bacterial virulence and survival post-infection.
• Identifying and mapping sRNA expression genome-wide, especially *de novo*, is challenging with existing high-throughput sequencing methods.
• Current methodologies often require multiple dependencies and lack targeted *de novo* sRNA identification approaches.

Purpose of the Study:

• To develop a novel, efficient, and targeted computational framework for the *de novo* identification of bacterial sRNAs.
• To create a user-friendly tool based on the Isolation Forest algorithm for sRNA discovery from sRNA-seq data.
• To comprehensively map and characterize sRNAs and small proteins in *Mycobacterium tuberculosis*.

Main Methods:

• Development of an Isolation Forest algorithm-based computational method for *de novo* sRNA identification.
• Implementation of the method into a publicly available tool: Prediction Of sRNAs using Isolation Forest (PoSIF).
• Application of the PoSIF tool to analyze bacterial sRNA-seq data, specifically from *Mycobacterium tuberculosis*.

Main Results:

• Successfully predicted 1120 small non-coding RNAs (sRNAs) and 46 small proteins in *Mycobacterium tuberculosis*.
• Demonstrated the capability of the PoSIF tool for *de novo* identification of sRNAs from sequencing data.
• Identified novel sRNAs with context-dependent expression, suggesting roles in stress response mechanisms.

Conclusions:

• The developed Isolation Forest-based method and PoSIF tool provide an effective approach for *de novo* bacterial sRNA identification.
• This study significantly expands the known repertoire of sRNAs and small proteins in *Mycobacterium tuberculosis*.
• The findings highlight the potential importance of novel sRNAs in bacterial adaptation and stress response.