Protocol to perform integrative analysis of high-dimensional single-cell multimodal data using an interpretable deep...

Area of Science:

• Single-cell biology
• Bioinformatics
• Computational biology

Background:

• Single-cell multi-omics sequencing allows leveraging multiple data types from individual cells.
• Analyzing high-dimensional multimodal single-cell data presents significant computational challenges.
• Existing methods may lack interpretability or struggle with integrating diverse omics layers.

Purpose of the Study:

• To present a protocol for the integrative analysis of high-dimensional single-cell multimodal data.
• To introduce moETM, an interpretable deep learning technique for multi-omics integration.
• To demonstrate the application of the protocol using bone marrow mononuclear cell data.

Main Methods:

• Development and application of the moETM deep learning framework.
• Protocol includes data preprocessing, multi-omics integration, and pathway knowledge incorporation.
• Cross-omics imputation strategies are detailed within the protocol.
• Demonstration using single-cell multi-omics data (GSE194122) from bone marrow mononuclear cells.

Main Results:

• The moETM protocol facilitates interpretable integration of single-cell multi-omics data.
• The method allows for the inclusion of prior biological pathway knowledge.
• Cross-omics imputation enhances data completeness and analytical power.
• Successful application demonstrated on a relevant biological dataset.

Conclusions:

• The presented protocol and moETM offer a powerful approach for single-cell multi-omics data analysis.
• This method enhances biological insights by integrating diverse omics modalities.
• The interpretability of moETM aids in understanding complex cellular heterogeneity.