Unico: a unified model for cell-type resolution genomics from heterogeneous omics data.

Area of Science:

• Genomics
• Computational Biology
• Bioinformatics

Background:

• Population-scale genomic datasets often comprise bulk samples from heterogeneous tissues.
• These bulk samples represent a mixture of various cell types, obscuring cell-specific biological insights.

Purpose of the Study:

• To introduce Unico, a novel unified cross-omics computational method.
• To deconvolve standard two-dimensional bulk matrices into three-dimensional tensors representing samples, features, and cell types.

Main Methods:

• Unico is a principled, model-based deconvolution approach.
• The method is theoretically justified for all tissue-level genomic data.
• Applied to bulk gene expression and DNA methylation datasets.

Main Results:

• Unico demonstrated superior performance compared to existing deconvolution methods.
• Successfully deconvolved bulk matrices into cell-type resolution data.
• Enhanced the capability for large-scale genomic studies.

Conclusions:

• Unico provides a robust framework for cell-type deconvolution across various genomic data types.
• Enables more powerful and precise analyses of complex biological tissues.
• Advances the field of single-cell resolution analysis from bulk samples.