Fig. 2: The performance comparison of the WES-based CNV callers on the 1000 Genomes test set which contains 157 samples. | Nature Communications

Fig. 2: The performance comparison of the WES-based CNV callers on the 1000 Genomes test set which contains 157 samples.

From: ECOLE: Learning to call copy number variants on whole exome sequencing data

Fig. 2

Left: Unpolished results vs. ECOLE and Right: DECoNT-polished results vs. ECOLE. CNVnator calls on the matched WGS samples are used as the ground truth. CNVkit and Control-FREEC return exact (integer) copy number predictions, which are discretized into deletion, duplication, and no-call. We also used the DECoNT tool to polish call sets of all considered tools which are denoted by DECONT-tool_name. Source data are provided as a Source Data file.

Back to article page