PyClone-VI: scalable inference of clonal population structures using whole genome data.

Area of Science:

• Computational biology
• Genomics
• Cancer research

Background:

• Tumors comprise diverse, genetically distinct malignant cell populations at diagnosis.
• Bulk sequencing and computational deconvolution identify these populations for cancer evolution studies.
• Current deconvolution methods struggle with speed and accuracy on large whole-genome sequencing datasets.

Purpose of the Study:

• To develop a computationally efficient Bayesian statistical method for inferring cancer clonal population structure.
• To provide an accurate and scalable tool for analyzing complex tumor genomic data.

Main Methods:

• Developed PyClone-VI, a Bayesian statistical approach for clonal deconvolution.
• Applied the method to large-scale cancer genomic datasets.

Main Results:

• PyClone-VI infers cancer clonal population structure efficiently.
• Demonstrated utility on 1717 patients from the PCAWG study and 100 from the TRACERx study.
• Achieved 10-100x speed improvement over existing methods with comparable accuracy.

Conclusions:

• PyClone-VI offers a significant advancement in computational cancer genomics.
• The method provides accurate and rapid analysis of tumor clonal architecture.
• Freely available software facilitates broader application in cancer research.