Skip to main content
Fig. 1 | BMC Cancer

Fig. 1

From: Bayesian copy number detection and association in large-scale studies

Fig. 1

Overview of sample processing, estimation of batch effects and copy number, and risk model for pancreatic cancer. a DNA samples for pancreatic cancer cases and healthy controls were obtained from 9 different study centers and processed centrally where samples were randomized to chemistry plates. b Initial preprocessing of these samples identified candidate CNV regions. As the principal sources of batch effects were unknown, we developed an approach to identify latent batch effects by clustering empirical cummulative distribution functions (eCDFs) of CNV region summaries (c) and to genotype these samples via a Bayesian hierarchical mixture model (d). Uncertainty of the copy number genotypes (e) was propagated from the genomic analyses to the Bayesian logistic regression model for pancreatic cancer risk (f)

Back to article page