Related Experiment Video
Updated: Aug 17, 2025
Droplet Barcoding-Based Single Cell Transcriptomics of Adult Mammalian Tissues
Published on: January 10, 2019
Annotating TSSs in Multiple Cell Types Based on DNA Sequence and RNA-seq Data via DeeReCT-TSS.
Juexiao Zhou1, Bin Zhang2, Haoyang Li2
1Computer Science Program, Computer, Electrical and Mathematical Sciences and Engineering Division, King Abdullah University of Science and Technology, Thuwal 23955-6900, Saudi Arabia; Computational Bioscience Research Center, King Abdullah University of Science and Technology, Thuwal 23955-6900, Saudi Arabia; Department of Biology, School of Life Sciences, Southern University of Science and Technology, Shenzhen 518055, China.
DeeReCT-TSS accurately identifies transcription start sites (TSSs) using DNA and RNA sequencing data. This deep learning method improves gene regulation understanding by precisely annotating TSS usage across cell types.
Area of Science:
- Genomics
- Molecular Biology
- Bioinformatics
Background:
- Accurate annotation of transcription start sites (TSSs) and their usage is crucial for understanding gene regulation.
- Existing computational tools for TSS prediction often yield high false positive rates due to their binary classification approach on balanced datasets.
- Genome-wide TSS identification requires robust methods that can handle the complexity of biological contexts.
Purpose of the Study:
- To develop a novel deep learning-based method, DeeReCT-TSS, for accurate genome-wide identification of transcription start sites (TSSs).
- To integrate both DNA sequence and RNA sequencing data for improved TSS prediction.
- To enable precise annotation of TSS usage across diverse cell types and identify cell type-specific TSSs.
Main Methods:
- Developed DeeReCT-TSS, a deep learning model integrating DNA sequence and RNA sequencing data for TSS identification.
- Implemented a meta-learning extension for simultaneous TSS annotation across 10 cell types.
- Validated the precision of predicted TSSs by correlating them with experimentally defined TSS chromatin states on independent datasets.
Main Results:
- DeeReCT-TSS significantly outperforms solely sequence-based methods in precise TSS annotation across different cell types.
- The meta-learning extension successfully identified cell type-specific TSSs.
- High precision of DeeReCT-TSS predictions was confirmed using chromatin state data from independent validation sets.
Conclusions:
- DeeReCT-TSS offers a powerful and precise approach for genome-wide TSS annotation by leveraging multi-modal data.
- The method enhances the understanding of gene regulation mechanisms through accurate TSS usage profiling.
- DeeReCT-TSS provides a valuable tool for genomic research, with accessible source code for the scientific community.
