Correction of transposase sequence bias in ATAC-seq data with rule ensemble modeling

Area of Science:

• Genomics
• Molecular Biology
• Bioinformatics

Background:

• Chromatin accessibility assays, such as ATAC-seq, are vital for studying gene transcription regulation.
• These assays provide single-nucleotide resolution of regulatory elements like promoters and transcription factor binding sites.
• A significant challenge is the inherent sequence bias of the Tn5 transposase enzyme used in ATAC-seq, which complicates data interpretation.

Purpose of the Study:

• To develop and validate a novel method for characterizing and correcting the complex sequence biases of the Tn5 transposase in ATAC-seq data.
• To enhance the accuracy of inferring transcription factor binding and regulatory element activity from chromatin accessibility measurements.

Main Methods:

• A rule ensemble machine learning approach was employed to model the Tn5 enzyme's sequence bias.
• The model integrates information from k-mers proximal to ATAC-seq reads to capture complex bias patterns.
• This method effectively characterizes both single-nucleotide and regional sequence biases.

Main Results:

• The developed machine learning model successfully characterized the intricate sequence biases of the Tn5 transposase.
• The method effectively corrected single-nucleotide and regional sequence biases present in ATAC-seq data.
• This bias correction improves the reliability of downstream analyses of chromatin accessibility.

Conclusions:

• Correcting Tn5 enzymatic sequence bias is crucial for accurate interpretation of ATAC-seq data.
• The machine learning approach offers a robust solution for addressing sequence bias in chromatin accessibility assays.
• Accurate bias correction advances the study of transcription regulation and genomic element activity.