De novo compartment deconvolution and weight estimation of tumor samples using DECODER.

Area of Science:

• Computational biology
• Genomics
• Cancer research

Background:

• Tumors are complex mixtures of neoplastic and non-neoplastic cells.
• Global gene expression analysis averages contributions, masking individual compartment profiles.
• Understanding non-neoplastic components is crucial for comprehensive cancer analysis.

Purpose of the Study:

• To develop an integrated computational framework for de novo deconvolution of tumor gene expression.
• To enable accurate estimation of single-sample compartment weights.
• To apply the framework to diverse cancer datasets and data types.

Main Methods:

• Developed DECODER, an integrated framework for deconvolution and compartment weight estimation.
• Applied DECODER to 33 The Cancer Genome Atlas (TCGA) tumor RNA-seq datasets.
• Validated DECODER's applicability to other data types, such as ATAC-seq.

Main Results:

• DECODER successfully deconvolves tumor RNA-seq data to estimate cellular compartment proportions.
• The framework reproducibly estimates clinically meaningful cellular compartment weights in pancreatic cancer.
• Demonstrated DECODER's utility across various cancer types and data modalities.

Conclusions:

• DECODER provides a robust method for identifying cellular compartments within complex tumor samples.
• This approach enhances the capability to analyze gene expression contributions from distinct cellular components.
• DECODER has potential implications for identifying the tissue of origin in cancers of unknown primary.