Integrating single-cell multimodal epigenomic data using 1D-convolutional neural networks.

Area of Science:

• Genomics
• Computational Biology
• Epigenetics

Background:

• Single-cell multimodal epigenomic profiling measures multiple histone modifications and chromatin accessibility simultaneously.
• Integrating these diverse epigenomic modalities is crucial for understanding cell-type-specific variations.
• Existing integration methods are not optimized for the unique sequential nature of this data.

Approach:

• Developed ConvNet-VAEs, a novel framework using 1D-convolutional variational autoencoders (VAEs).
• Models single-cell multimodal epigenome data as a multi-channel sequential signal.
• Evaluated on nano-CT and scNTT-seq data from mouse brain and human bone marrow.

Key Points:

• ConvNet-VAEs achieve superior dimension reduction and batch correction compared to previous architectures.
• The framework uses significantly fewer parameters than alternative methods.
• Convolutional architectures show improved performance with increasing modalities, unlike fully-connected networks.

Conclusions:

• ConvNet-VAEs offer a promising approach for integrating single-cell multimodal epigenomic data.
• Convolutional autoencoders are well-suited for current and future epigenomic datasets.
• This method advances the analysis of complex epigenomic landscapes.