Skip to main content
Fig. 3 | BMC Medical Genomics

Fig. 3

From: Clinical analysis of germline copy number variation in DMD using a non-conjugate hierarchical Bayesian model

Fig. 3

Classification performance with increasing read pair coverage Individual subject target intensities for nine simulated subjects were generated from hyperparameters estimated from a cohort of 38 volunteer subjects. True copy number states from nine Coriell test subjects (eight with CNVs and one negative control) were used to set multinomial probabilities before read pair coverage simulation. Panel x-axes indicate the total number of simulated read pairs mapping to the relevant exon targets (including the baseline targets). a and b indicate classification performance under the credible interval cutoffs of 0.99 and 0.9 respectively (i.e. targets where the highest-density interval of the chosen size overlaps two copy number states are not assigned a call). Callable targets are those assigned a final copy number call given the chosen cutoff. c and d display the copy number state visualization produced after MCMC sampling c indicates a typical result using a low read pair coverage (750 total read pairs). The underlying copy number states are unidentifiable. d shows results for a simulated sample with the same true copy number states as c but a total read pair coverage of 45,000 (approximately 20,700 at the targets of interest)

Back to article page