Related Experiment Video
Updated: Sep 2, 2025
10:00
An Ultrahigh-throughput Microfluidic Platform for Single-cell Genome Sequencing
Published on: May 23, 2018
17.8K
scBasset: sequence-based modeling of single-cell ATAC-seq using convolutional neural networks.
1Calico Life Sciences, South San Francisco, CA, USA. yuanh@calicolabs.com.
Nature Methods
|August 8, 2022
Summary
scBasset, a novel deep learning method, enhances the analysis of single-cell ATAC sequencing (scATAC) data by utilizing DNA sequence information. This approach improves cell clustering, data denoising, and transcription factor activity inference for epigenetic studies.
Area of Science:
- Genomics
- Computational Biology
- Epigenetics
Background:
- Single-cell assay for transposase-accessible chromatin using sequencing (scATAC) is crucial for understanding cellular heterogeneity in epigenetic landscapes.
- Analysis of scATAC data faces challenges due to high dimensionality and data sparsity.
Purpose of the Study:
- Introduce scBasset, a sequence-based convolutional neural network, to effectively model scATAC data.
- Improve the analysis of single-cell epigenetic datasets.
Main Methods:
- Developed scBasset, a deep learning model leveraging DNA sequence information underlying chromatin accessibility peaks.
- Applied scBasset to various scATAC and single-cell multiome datasets.
Main Results:
- scBasset achieved state-of-the-art performance in multiple analytical tasks.
- Demonstrated improved cell clustering, scATAC profile denoising, and data integration.
- Showcased accurate transcription factor activity inference.
Conclusions:
- scBasset offers a powerful new approach for analyzing complex single-cell epigenetic data.
- Leveraging DNA sequence and neural networks significantly advances scATAC data interpretation.
