Inference of clonal selection in cancer populations using single-cell sequencing data

Abstract

Summary: Intra-tumor heterogeneity is one of the major factors influencing cancer progression and treatment outcome. However, evolutionary dynamics of cancer clone populations remain poorly understood. Quantification of clonal selection and inference of fitness landscapes of tumors is a key step to understanding evolutionary mechanisms driving cancer. These problems could be addressed using single-cell sequencing (scSeq), which provides an unprecedented insight into intra-tumor heterogeneity allowing to study and quantify selective advantages of individual clones. Here, we present Single Cell Inference of FItness Landscape (SCIFIL), a computational tool for inference of fitness landscapes of heterogeneous cancer clone populations from scSeq data. SCIFIL allows to estimate maximum likelihood fitnesses of clone variants, measure their selective advantages and order of appearance by fitting an evolutionary model into the tumor phylogeny. We demonstrate the accuracy our approach, and show how it could be applied to experimental tumor data to study clonal selection and infer evolutionary history. SCIFIL can be used to provide new insight into the evolutionary dynamics of cancer.

Availability and implementation: Its source code is available at https://github.com/compbel/SCIFIL.

Fig. 1.

Mutation tree

Fig. 2.

Depiction of the evolutionary model. Tree nodes represent mutation events whose times are marked on the time axis. Leafs represent the sampling event. For each node, the distribution of clone abundances after the corresponding event is shown

Fig. 3.

Example of simulated mutation tree

Fig. 4.

Performance of SCIFIL on simulated data with m mutations and fixed standard deviation of mutation rate. (Left): Mean relative accuracy of fitness estimation. (Right): Spearman correlation between true and inferred fitness vectors

Fig. 5.

Performance of SCIFIL on simulated data with m = 50 mutations and different standard deviations of mutations rates. (Left) Mean relative accuracy of fitness estimation. (Right) Spearman correlation between true and inferred fitness vectors

Fig. 6.

Performance of SCIFIL on simulated data with different false negative error rates α and with mutation trees reconstructed by SCITE (Jahn et al., 2016). (Left) Mean relative accuracy of fitness estimation. (Right) Spearman correlation between true and inferred fitness vectors

Fig. 7.

(Left) Spearman correlation between true and inferred fitness vectors for QuasiFit and SCIFIL. (Right) Running time of SCIFIL

Fig. 8.

Fitness landscape and mutation tree for JAK2-negative myeloproliferative neoplasm (Hou et al., 2012) (left) and colorectal cancer (right) (Leung et al., 2017) inferred by SCIFIL. Colors represent fitness values and distance from each tree node to the root is approximately proportional to its time of appearance

Fig. 9.

Log-likelihoods of trees with and without recurrent mutations. (Left) Log-likelihoods produced by infSCITE. (Right) Evolutionary likelihoods produced by SCIFIL. Likelihoods of perfect phylogeny are shown in green. Purple and red: trees with the evolutionary likelihoods higher than for the perfect phylogeny