PureCN: copy number calling and SNV classification using targeted short read sequencing

Abstract

BackgroundMatched sequencing of both tumor and normal tissue is routinely used to classify variants of uncertain significance (VUS) into somatic vs. germline. However, assays used in molecular diagnostics focus on known somatic alterations in cancer genes and often only sequence tumors. Therefore, an algorithm that reliably classifies variants would be helpful for retrospective exploratory analyses. Contamination of tumor samples with normal cells results in differences in expected allelic fractions of germline and somatic variants, which can be exploited to accurately infer genotypes after adjusting for local copy number. However, existing algorithms for determining tumor purity, ploidy and copy number are not designed for unmatched short read sequencing data.ResultsWe describe a methodology and corresponding open source software for estimating tumor purity, copy number, loss of heterozygosity (LOH), and contamination, and for classification of single nucleotide variants (SNVs) by somatic status and clonality. This R package, PureCN, is optimized for targeted short read sequencing data, integrates well with standard somatic variant detection pipelines, and has support for matched and unmatched tumor samples. Accuracy is demonstrated on simulated data and on real whole exome sequencing data.ConclusionsOur algorithm provides accurate estimates of tumor purity and ploidy, even if matched normal samples are not available. This in turn allows accurate classification of SNVs. The software is provided as open source (Artistic License 2.0) R/Bioconductor package PureCN (http://bioconductor.org/packages/PureCN/).Electronic supplementary materialThe online version of this article (doi:10.1186/s13029-016-0060-z) contains supplementary material, which is available to authorized users.

Tracking the Evolution of Resistance to ALK Tyrosine Kinase Inhibitors Through Longitudinal Analysis of Circulating Tumor DNA

“…Variants were called with Pindel, Socrates, PureCN, and MuTect using the default 0.5% lower threshold. 19–22 Variants were then annotated, restricted to non-synonymous mutations involving protein-coding regions, and filtered using common reference databases and internal controls to remove germline variants and artifacts. All ALK resistance mutations were re-genotyped with Genome Analysis Toolkit.…”

Optimized data representation and convolutional neural network model for predicting tumor purity

“…We then sought to build a model to predict purity by learning the statistics of the input data, as opposed to hand-crafting a statistical model, as with previous methods [10,11]. Additionally, given that expert human annotators use scatterplots to visualize the data, we reasoned that the very same scatterplots would and should serve as the best data representation for our model.…”

“…We built and trained a CNN regression model to output tumor purity based upon the very scatterplot images of input data that an expert human annotator would use for prediction. We found that with our data representation, combined with a CNN and data augmentation, we could predict purity better than an existing algorithm [10].…”

“…tumor ploidy) and tumor/normal cell proportion (i.e. tumor purity) that best fits the observed variant allele frequency [10,11]. These automated methods, although powerful, still require manual inspection of results, as they may yield results inconsistent with those generated by an expert human annotator.…”

“…Specifically, these methods may incorrectly estimate tumor ploidy and purity, especially with noisy samples, in many cases where a human annotator might easily make a correct estimation. Lastly, only one of these methods, PureCN [10], is a fully open-source, non-commercial product available for general use.…”