Skip to main content
Fig. 3 | BMC Bioinformatics

Fig. 3

From: Modeling and correct the GC bias of tumor and normal WGS data for SCNA based tumor subclonal population inferring

Fig. 3

GC bias of WGS data of tumor-normal paired sample HCC1954.mix1.n20t80 of TCGA mutation calling benchmark 4. Let DS and DN respectively denote the read counts of the segment of tumor and normal samples. a The GC bias of the Log ratio of tumor and normal read counts of the SCNA segments. The purple and blue lines are linear regression and loess regression lines respectively. b The GC bias of the ratio of tumor and normal read counts of the SCNA segments. The red line are drawn by the loess regression model with a quadratic polynomial function, which is used to rectify the distribution of the ratio DS/DN in the state-of-art GC correction method [14]. c The GC bias of the ratio of tumor and normal read counts of the 5000 bp bin. Since the majority (81%) of CNV calls are between 1 kb and 100 kb [17], most of 5000 bp bins spans only one SCNA. This sub-figure shows most SCNAs clustered clearly into multiple strips. d The GC bias of the ratio of tumor and normal read counts of the 500 bp bin. e The distribution of B-allele frequency (BAF) of stripe 1–6 in Fig. 3a. The SCNA segments are obtained by BIC-seq [18]

Back to article page