Skip to main content
Fig. 3 | BMC Cancer

Fig. 3

From: A machine learning approach to optimizing cell-free DNA sequencing panels: with an application to prostate cancer

Fig. 3

Panel performance using in silico capture of cfDNA. Five patients with multiple prostate cancer tumor foci and normal tissue DNA were whole exome sequenced at 200X-fold coverage. Discovered somatic variants were in silico “captured” with three panels: 1) our orchid generated panel, 2) a panel consisting of all mutations in the ICGC prostate cancer dataset with a frequency > 1 patient, and 3) a panel consisting of genes on any of 4 clinically used panels (union-existing). The mean number of total somatic mutations across foci are listed below each patient and the mean numbers of those present on each of the three panels are shown (blue bars). Orchid detected significantly more mutations in all patients except P0024 (with only one focus; union-existing [p < 0.03], frequency [p < 0.02] using a T-test)

Back to article page