Combining signal and sequence to detect RNA polymerase initiation in ATAC-seq data

Area of Science:

• Genomics
• Molecular Biology
• Bioinformatics

Background:

• Assay for transposase-accessible chromatin followed by sequencing (ATAC-seq) is a cost-effective and low-cell-input method for mapping open chromatin regions.
• Its simplicity makes it suitable for various research settings where other high-throughput assays are difficult to implement.
• Understanding the full potential and limitations of ATAC-seq data is crucial for accurate biological inference.

Purpose of the Study:

• To investigate the capability of ATAC-seq data in predicting nascent transcription, a key indicator of active gene expression.
• To establish a method for identifying genome-wide active transcription regions using ATAC-seq.
• To assess the performance of this method across different cell types.

Main Methods:

• Leveraging ATAC-seq data by combining mapped short reads with underlying peak sequences.
• Developing a hybrid signal/sequence representation of the ATAC-seq data.
• Utilizing recurrent neural networks (RNNs) for classification to identify regions of active transcription.

Main Results:

• Demonstrated that ATAC-seq data, when analyzed with a hybrid signal/sequence approach and RNNs, can accurately predict nascent transcription.
• Successfully identified genome-wide regions of active transcription across diverse cell types.
• Showcased the potential of ATAC-seq as a surrogate for nascent transcription assays.

Conclusions:

• ATAC-seq is a versatile tool that can be extended to infer transcriptional activity beyond just chromatin accessibility.
• The developed RNN-based method provides a robust way to detect nascent transcription from ATAC-seq data.
• This approach enhances the utility of ATAC-seq for studying gene regulation and functional genomics.