Revisiting the cancer microbiome using PRISM.

Area of Science:

• Genomics
• Microbiology
• Bioinformatics

Background:

• The cancer microbiome's role is debated, necessitating advanced microbial analysis for human genomics.
• Accurate identification of microorganisms in low-biomass samples is crucial for cancer research.

Purpose of the Study:

• To develop and validate PRISM, a computational method for precise microorganism identification and decontamination in low-biomass sequencing data.
• To apply PRISM to analyze tumor-associated microbes across various cancer types using CPTAC and TCGA datasets.

Main Methods:

• PRISM computational approach for microorganism identification and decontamination.
• Benchmarking PRISM on a large curated dataset (62,006 taxa).
• Analysis of 8 cancer types from CPTAC and TCGA datasets.

Main Results:

• PRISM demonstrates excellent performance in identifying true and false-positive taxa.
• Rich microbiomes detected in gastrointestinal tumors (CPTAC).
• Bacteria identified in pancreatic tumors associated with altered glycoproteomes, smoking history, and recurrence risk.
• Sparse microbial signals in other cancer types and TCGA data, potentially due to sequencing parameters.

Conclusions:

• PRISM enhances confidence in microbial analyses of human genomic data.
• Identified tumor-associated microorganisms possess potential molecular and clinical significance.
• The study highlights PRISM's utility in uncovering microbial roles in cancer, particularly pancreatic cancer.