Skip to main content
Fig. 6 | Genome Biology

Fig. 6

From: DREAMS: deep read-level error model for sequencing data applied to low-frequency variant calling and circulating tumor DNA detection

Fig. 6

Prediction of cancer using a DREAMS-cc, b Shearwater, and c Mutect2. For each patient’s LB-sample (y-axis), the mutation catalog (x-axis) for every candidate patient is used for calling cancer. The patients are stratified into patients with stage I and stage II CRC. The diagonal is showing the result of using a patient’s own mutation catalog for cancer calling and constitutes the expected positives. The off-diagonal is the cross-patient results, for which the mutation catalog is filtered with the patient’s tumor and germline variants prior to cancer calling, and thus these are expected to be negative. The color scheme is chosen based on the matched quantiles of the rank from the p-value, combined Bayes factors and max TLOD from a DREAMS-cc, b Shearwater, and c Mutect2, respectively. The cancer predictions show the results from one split in the 5 × 2 CV. d AUC performance of DREAMS-cc, shearwater, and Mutect2 with respect to calling cancer using the 5 × 2 CV

Back to article page