Related Experiment Video
Updated: May 23, 2025
08:01
Author Spotlight: Unraveling Bacterial Responses to Antibiotics and Immune System in Tissues
Published on: March 1, 2024
805
BacTermFinder: a comprehensive and general bacterial terminator finder using a CNN ensemble
Seyed Mohammad Amin Taheri Ghahfarokhi1, Lourdes Peña-Castillo1,2
1Department of Computer Science, Memorial University of Newfoundland, St. John's, Newfoundland A1B 3X5, Canada.
NAR Genomics and Bioinformatics
|March 10, 2025
Summary
BacTermFinder, a new tool using convolutional neural networks (CNNs), accurately predicts bacterial transcription terminators. It identifies both intrinsic and factor-dependent types, outperforming existing methods and generalizing to archaea.
Area of Science:
- Genomics
- Bioinformatics
- Computational Biology
Background:
- Transcription terminators are crucial DNA regions that signal the end of gene transcription.
- Existing computational tools for predicting bacterial terminators often lack specificity, being limited to certain bacterial species or terminator types (intrinsic or factor-dependent).
Purpose of the Study:
- To develop a novel, accurate, and broadly applicable computational tool for predicting bacterial transcription terminators.
- To address the limitations of existing methods by creating a tool that can identify both intrinsic and factor-dependent terminators across diverse bacterial species and generalize to archaea.
Main Methods:
- Development of BacTermFinder, an ensemble of convolutional neural networks (CNNs).
- Training the model on approximately 41,000 bacterial terminators (intrinsic and factor-dependent) from 22 species with diverse GC content (28%-71%), sourced from RNA-seq studies.
- Inputting four distinct representations of terminator sequences into the CNNs.
Main Results:
- BacTermFinder demonstrated superior performance compared to four other bacterial terminator prediction tools, achieving higher average recall without an increase in false positives.
- The tool successfully identified both intrinsic and factor-dependent terminator types.
- BacTermFinder exhibited generalization capabilities, accurately predicting archaeal terminators.
- Saliency maps were utilized to visualize CNN insights into species-specific terminator motifs.
Conclusions:
- BacTermFinder represents a significant advancement in computational prediction of bacterial transcription terminators, offering improved accuracy and broader applicability.
- The tool's ability to identify diverse terminator types and generalize to archaea enhances its utility in genomic analysis.
- The visualization of CNN saliency maps provides valuable insights into the sequence features governing termination across different species.
