Inferring Markov Chains to Describe Convergent Tumor Evolution With CIMICE

Area of Science:

• Computational Biology
• Genomics
• Cancer Research

Background:

• Tumor phylogenetics is crucial for understanding cancer cell heterogeneity.
• Existing cancer progression models are data-dependent and limited by technological assumptions.
• Evolving experimental technologies necessitate tailored modeling approaches.

Purpose of the Study:

• To develop a cancer progression model specifically for Single Cell DNA Sequencing (scDNA-seq) data.
• To create a flexible Directed Acyclic Graph (DAG) model capable of identifying progression beyond the infinite site assumption.
• To provide a conservative modeling framework that avoids inferring unrepresented knowledge.

Main Methods:

• Defining a minimal set of assumptions for DAG-based model reconstruction.
• Tailoring the modeling formalism to the specific characteristics of scDNA-seq data.
• Developing an open-source R implementation named CIMICE, available on BioConductor.

Main Results:

• Demonstrated the model's features through simulations and analytical results.
• Validated the model's performance on real cancer data.
• Showcased the model's integration capabilities with other methods to handle input noise.
• Developed a framework for generating simulated data consistent with theoretical assumptions.

Conclusions:

• The proposed CIMICE model offers a flexible and data-specific approach to tumor phylogenetics using scDNA-seq.
• The model successfully identifies cancer progression beyond the infinite site assumption.
• CIMICE provides a valuable, open-source tool for cancer research and data simulation.