Power and pitfalls of computational methods for inferring clone phylogenies and mutation orders from bulk sequencing data

摘要: Tumors harbor extensive genetic heterogeneity in the form of distinct clone genotypes that arise over time and across different tissues regions cancer. Many computational methods produce phylogenies from population bulk sequencing data collected multiple tumor samples a patient. These are used to infer mutation order origins during progression, rendering selection appropriate clonal deconvolution method critical. Surprisingly, absolute relative accuracies these correctly inferring yet consistently assessed. Therefore, we evaluated performance seven methods. The accuracy reconstructed inferred groupings varied extensively among All tested showed limited ability identify ancestral sequences present correctly. presence copy number alterations, occurrence seeding events sites metastatic evolution, intermixture cancer cells tumors hindered detection clones inference for all tested. Overall, CloneFinder, MACHINA, LICHeE highest overall accuracy, but none performed well simulated datasets. So, guidelines selecting analysis.