Reliable detection of Host-Microbe Signatures in cancer using PRISM.

Area of Science:

• Computational biology
• Microbiome research
• Cancer genomics

Background:

• The cancer microbiome field faces challenges with reliable microbial detection from human genomic data.
• Low-biomass samples often yield ambiguous results, leading to controversy.
• Accurate identification of microorganisms is crucial for understanding cancer development and treatment.

Purpose of the Study:

• To develop an efficient computational framework, PRISM, for precise microorganism identification and decontamination from low-biomass sequencing data.
• To benchmark PRISM's performance on diverse datasets.
• To apply PRISM to analyze cancer microbiome signatures across various cancer types.

Main Methods:

• Development of PRISM, an efficient computational framework for microbial identification and decontamination.
• Benchmarking PRISM on 230 independent datasets with known microbial taxa.
• Profiling 25 cancer types using The Cancer Genome Atlas and Clinical Proteomic Tumor Analysis Consortium data.

Main Results:

• PRISM demonstrated robust performance in microbial detection and decontamination.
• Consistent microbial signatures were identified in gastrointestinal, head-and-neck, and urogenital tract tumors.
• In pancreatic cancer, microbial detection correlated with altered host protein glycosylation and smoking exposure.

Conclusions:

• PRISM significantly improves the reliability of microbial profiling from human genomic data.
• The framework enables leveraging existing genomic data for concurrent host-microbial signature detection.
• Findings suggest potential molecular and clinical significance of host-microbial interactions in cancer.