Inference of tumor phylogenies from genomic assays on heterogeneous samples

摘要: Motivation: Tumorigenesis can in principle result from many combinations of mutations, but only a few roughly equivalent sequences or "progression pathways", seem to account for most human tumors. There is hope that by cataloguing the common progression pathways, we identify broadly useful therapeutic targets and reliable diagnostic tests determine who will benefit any given treatment. Phylogenetic approaches, which rely on observation tumors are evolving populations cells, provide promising way robust evolutionary features across They face their own challenges, though, including high heterogeneity individual tumors, makes it difficult specific steps progression.Results: We previously developed algorithms inferring cell states heterogeneous tumor samples. Here, build work develop pipeline phylogenetic inference these inferred states. statistical method identifying differentially amplified chromosome regions show application breast cancer comparative genomic hybridization (CGH) data set effective at biologically meaningful, phylogenetically informative markers. then perform phylogeny resulting marker construct tree several major pathways predicting possible ancestral states.Conclusions: This demonstrates feasibility unmixing methods facilitating In process, provides predictions key development may be new subtypes markers ultimately prove novel diagnostics therapeutics.Availability: New code this paper made available interested researchers upon request.