TRONCO

TRONCO collects algorithms to infer cancer progression models via Suppes-Bayes Causal Network from cross-sectional, multi-region, or single-cell tumor samples.

Key Features:

• Progression Model Inference: Implements Suppes-Bayes Causal Network algorithms in parallel to infer progression models from binary alteration matrices.
• Multi-sample Type Support: Processes both ensemble tumor data (cross-sectional samples) and individual patient data (multi-region or single-cell samples).
• Data Import: Imports data from plain files, MAF files, GISTIC format files, and fetches data from cBioPortal for cancer genomics.
• Data Exchange: Provides import/export functions for interoperability with other bioinformatics tools for clustering or detection of mutually exclusive alterations.
• Model Validation: Supports bootstrap and cross-validation to test confidence of inferred models.
• Data Manipulation and Visualization: Includes functions for manipulating and visualizing data and inferred models.

Scientific Applications:

• Cancer Progression Modeling: Reconstructs temporal sequences of somatic alterations driving tumor evolution in cancer genomics studies.
• Intra-tumor Heterogeneity Analysis: Models clonal evolution within individual patients using multi-region or single-cell sequencing data.
• Mutual Exclusivity Detection: Facilitates identification of mutually exclusive genomic alterations through data export to specialized tools.
• Pipeline for Cancer Inference (PICNIC): Serves as the implementation platform for the PICNIC methodology.

Methodology:

TRONCO processes binary matrices where rows represent tumor samples and columns represent single-nucleotide or structural variants, with 0/1 values indicating absence or presence of alterations. The Suppes-Bayes Causal Network approach is applied to infer progression models, with parallel algorithm implementation for computational efficiency.